Forrest,

You are right, we (or better to say: myself) might have mentioned that Kevin

Peter Enneson's essay is brilliant.

Everything I have always wanted to say about the subject (but could never do so half as well).

The brightness contrast of some of the text on color backgrounds surprises me for an issue on readability. The first page of Hrant's article's black on dark ochre is tough enough but the green-on-green of Enneson's article was really disruptive even for such a well written piece. The Dyson article used a raster pattern about half way down the page making that portion of text all but unreadable to me.
All-in-all it is a great issue for content and I commend TYPO for such a wonderful editorial interplay of ideas.
It makes me wonder if they were pushing the boundaries of readability with color and pattern on purpose just to see how far they could go before someone said "ouch."

ChrisL

Chris, after reading text on the IntarWEB for the last 10 years, I'm grateful when the letterforms are legible. I honestly didn't even notice the color contrasts (or lack thereof) when I was reading. It all seemed perfectly readable to me!

And Pavel, no apologies are necessary; Keith's paper deserves wider distribution and it gives Hrant something to work against. The only reason I was bummed was because there wasn't more new TYPO for me to read!

Am I the only person who finds reading white on black screen text easier than black on white? It would probably be different if I weren't using a CRT monitor, but solid white seems to hurts my eyes.

Black on off-white seems to be best. Not sure if this would be true for ClearType rendering on an LCD.

http://psychology.wichita.edu/optimalweb/text.htm

>> As for color, as long as there is sufficient contrast between the text and the background, many color combinations are possible. However, most studies have shown that dark characters on a light background are superior to light characters on a dark background (when the refresh rate is fairly high). For example, Bauer and Cavonius (1980) found that participants were 26% more accurate in reading text when they read it with dark characters on a light background. Moreover, a survey by Scharff, et al. (1996) revealed that the color combination perceived as being most readable is the traditional black text on white background. However, it is common for websites (such as this one) to have an off-white background in order to reduce the flicker and glare associated with white backgrounds. <<

I was reading the actual offset press printed magazine, not the PDF on screen and my 61 year old eyes could not finish an article at a sitting. I was very interested in the subject matter so I persisted and read it in several sessions.

>solid white seems to hurts my eyes.

Try turning down the brightness, especially on LCD.

My monitor is a CRT and it is adjusted fairly dim. I guess white doesn't really hurt my eyes, but it is more difficult to look at.

I just feel like the pure white background (made by additive light) are much more taxing to read upon than a black background. I've had a number of conversations about it now and I think it's weird that I can't find anyone sympathetic. For a while I thought it might be because my first 5 years on a computer were spent entirely in DOS with a black background.

I'm completely comfortable with being wrong on the subject, but I think there is more to it than that. Oh well. I hope the issue will go away as soon as I can afford a flat panel.

I've read all three articles, or rather did my first run- through.

Forrest wrote: "His language is occasionally arcane" in regards to Peter Enneson's article. I found it rather dense, and strewn with vocabulary that will deter many. I much prefer Kevin Larson's clear, expository style among the three. I was hoping to refer working, less-academically inclined, designers to the issue, but now I've got serious reservations. I'm not looking for a predigested or dumbed-down version, but in all fairness, a less erudite and more straightforward discussion by Enneson would help draw attention to the substantive issues here.

Speaking as a working designer, not academically inclined (and not inclined to the computer lab either), I think Peter Enneson's article is bang on. Yes, it's difficult, and some of his paragraphs have to be read several times, but this is a complex subject, and the humanist, intellectual approach does it more justice than the grandiose, simplistic notion of "The Science of Word Recognition" (Lorne Greene voice).

Wouldn't it be nice if there were some easy science that we could learn about reading, so that we could be better type designers and typographers? No, because once it's reduced to easy science, the job will be automated.

I'm not quite certain how to 'solve' the compact and perhaps somewhat distended expository style of my Typo 13 contribution except to say that, in the context of Hrant's somewhat cavalier or opportunistic approach to questions of evidence, and Kevin's failure to address matters of construct validity, I thought it necessary to develop the kind of text I did. I've asked Pavel if it might be possible to provide and English-language translation of the Czech-language summary supplied by David Barton. Some of my more arcane usages have to do with my sense that the typographical craft-evolved terminological repertoire might not be adequate to the task of articulating a typographically relevant evidence-based understanding of reading.

Another solution might be to turn my contribution into a much larger text, but at the moment my construction of reading is still too exploratory and my command of the relevant literature still too fragmentary to make that possible. My text as it stands represents a, for me, foundational document on which, I hope, I and others can build. But a larger text might be even less approachable for the graphic design student who wants an overview of the issues more than a detailed elaboration and justification of points of view.

In terms of substantive issues, let me offer the following: contra Kevin, I argue that the parallel letter recognition model of word recognition is neither sustainable in broadly cognitive-scientific terms nor convenient for gauging and channeling typographic practice; against Hrant I argue that the importance of internal features discrimination to the wordform (or in Hrant's terms: bouma) resolution mechanisms underlying immersive reading does not warrant a parafoveally skewed understanding of reading of the kind Hrant puts forward, and hence motivates a somewhat different understanding than his of what counts in typographic practice.

I also wanted to convey that the key relevant issues for typographic practice are 'perceptual processing' issues and that the benefit of addressing such issues at a fairly sophisticated level lies not in telling us in specifics what to do, but in identifying and orienting us toward what's important in typographic formgiving and typeform construction. (For example, one consequence of my outlook is to suggest that when using or designing a sans-serif font, inter- and intra-letter spacing--managing the white of the word--are more critical than they perhaps might be in old-style seriffed fonts.)

>against Hrant I argue that the importance of internal features discrimination to the wordform (or in Hrant's terms: bouma) resolution mechanisms underlying immersive reading does not warrant a parafoveally skewed understanding of reading of the kind Hrant puts forward, and hence motivates a somewhat different understanding than his of what counts in typographic practice.

Mr. Enneson, in my opinion this, and much in your essay, is bad writing--correction, terrible writing. There is no call for it in your subject matter. All you want to say, you can say with direct, clear language.

Joseph M. Williams, in his admirable book Style: Toward Clarity and Grace characterizes the way you are writing here as 'useless nominalization.'

Here is a comparative example from Williams:

"1a. Because we knew nothing about local conditions, we could not determine how effectively the committee had allocated funds to areas that most needed assistance.

"1b. Our lack of knowledge about local conditions precluded determination of committee action effectiveness in fund allocation to those areas in greatest need of assistance."

These say exactly the same thing. The first is readable, if a bit complex. The second is a puzzle that you have to unscramble. Your essay follows the second style, to the extreme.

I hope that when you write further on these subjects, you follow Williams' good counsel.

William, I could have said (had I taken more time and had I had less of a concern with putting the whole of my concerns into one comprehensive sentence):
In Hrant's understanding of reading parafoveal processing does the bulk of the work. Converging lines of evidence suggest that an understanding of reading such as this is not sustainable. The lines of evidence I have in mind relate to: 1) the role actually played in recognition by internal features of boumas; 2) the parafovea's inability to discriminate internal features effectively in bouma-sized contexts; and 3) recent denominations of the parafoveal preview benefit. This motivates a somewhat different understanding than his of what counts in typographic practice.

Or simply: studies show Hrant's understanding of reading is not sustainable.

I am not convinced I am guilty of 'useless nominalization'. 'Nominalization,' perhaps; 'useless,' I want to think, not. But, having said that, I am bothered by my lack of ability to communicate effectively to a sizable portion of the typographical community (of which I am a part) things I think need to be addressed; or to put it differently, I am disturbed by my ability to obscure what I think needs saying by provoking negative reactions to how I say it, or how I think it is best formulated. I hope readers will bear with me.

This was a double post of the previous entry--my apologies. Can it be deleted?

Ce qui se con

> In Hrant's understanding of reading parafoveal
> processing does the bulk of the work.

Real quick:
The most I've ever thought the parafovea (or really, more generally and significantly, multi-letter boumas) contributes is about 2/3-rds. But since Kevin's arrival I've downgraded that. However, I still think it's more than half. But anyway, really, the important thing isn't how much it contributes, but that it exists (something the LP model refutes entirely). I'm sorry if this is all too fuzzy to build an empire on, but please don't shoot the messenger.

More in time.

hhp

Peter, I am primarily a writer, and recently a typogrographer. I write on difficult subjects, and labor very long and hard to make myself understood. I hope you understand that it is because I think you have something to say that I was so brutally blunt. And I appreciate your gracious reply.

You show by your re-write that you are perfectly capable of putting in the labor on clear writing and getting a far better result.

You say that you are bothered by your difficulty in making yourself understood to typographers. Writing experts, in particular Mr. Williams, have spent their careers trying to identify precisely what makes writing more and less readable. I would suggest that their study and expertise should be taken as seriously as studies on what make type physically readable.

If you are concerned to communicate effectively, I would invite you to study Mr. Williams' book and judge for yourself whether his standards are good, and whether your nominalizations are useless or not. Mr. Williams discusses the difference between useful and useless nominalizations at length.

My own prediction is that if you try to follow his recommendations you will find yourself phenomenally better understood. I find that they do require a lot of work, but always get a better result.

Finally, on Yves' aphorism. I think it also works the other way around. When you try to make yourself clear by standards of good rhetoric, it helps you to clarify your own ideas. So the benefits is double: to you and to the reader.

To start with, you will find that the goal of putting everything into one sentence is a prescription for writing disaster.

Best of luck on this other, critical dimension of readability!

> useful and useless

The fatal flaw in this -which I see paralleled in the blind lust for totally pinning down reading among some people- is that different people find different things "useful". Peter's writing is not useless simply because it's difficult. Not everything that needs to be said can be said succinctly. Look at the most compelling examples of world literature - much of it would be gobbledygook by this standard. But it's not - humans love it, cherish it, benefit from it. You might say this is a technical subject that can't have a useful "existentialim". I would counter that this stance is cozy escapism, that everything a human can and does talk about has all the dimensions in it to some extent. There is a place in this dialog for all kinds of methods of expression, far beyond what can supposedly be "expositioned" in labs, especially since we're currently unable -and I feel will be unable for the forseeable future- to fully understand reading. You don't marry somebody you fully know.

BTW, I think you misunderstood Yves (unless I have). But of him I would ask: Et que doit on faire au sujet des choses qui ne se con

>Peter's writing is not useless simply because it's difficult

Hrant, I didn't say Peter's writing is useless.

I said Peter wrote with useless nominalization. In Williams' view a nominalization is useless when it does not help the writer express clearly what he or she wants to say. Instead it usually puts an obstacle in the reader's way.

Peter's was the only article in the magazine that I read in full. But then, I am a junkie for big words.

> In Williams' view a nominalization is useless when ....

To me this is exactly the sort of formalistic compartmentalization that precludes about half of humanly useful communication.

> Peter's was the only article in the magazine that I read in full.

For those who have already made up their minds (like Forrest too), that's certainly a smart time-saver. The important thing is the heady euphony of it all, of course - nevermind the content. Nevermind that Peter and I are much close than either of us to Kevin, but some people still gayly choose to favor the style of one and the implications of another, but of course not either enough to do something about it.

--

It's the hybrid mongrel, the fuzzy stuff in the middle that's most useful, as flightly and stinky as it is. This is what started to be lost about two thousand years ago, and has now become an inconvenient ghost of a memory for mankind.

hhp

For those who have already made up their minds (like Forrest too), that's certainly a smart time-saver. The important thing is the heady euphony of it all, of course - nevermind the content.

Hey now, I read all three articles

What's woolly? It seems to me that you're dismissing what you can't put in a box. The old "what can't be counted doesn't count". Is the entire corpus of anecdotal evidence woolly? Is my pointing out that the PL model can't properly explain very long saccades, regressions and typos woolly? Your own admission that we don't really know what's going on should make you more open to delve deeper, to wonder, to doubt - to avoid the succubus of the Easy Answer.

And you can think about and discuss style, or what kind of tea you like (I like Lapsang Souchong, and I'm OK with the fact that it makes most people nauseous), but don't let that cloud the evaluation of the content. None of the stuff in Typo13 is bullshit. It's all what each of us is thinking, inescapably to some extent feeling. Humans do this.

hhp

I'm not eager to draw out discussion of my rhetorical strategies (or those of others) to the point where discussion of the issues, and their meaning for practice is indefinitely postponed. But let me add, William, that I'm not entirely convinced that texts like the one I prepared for publication don't have a value and a place. In preparing the text I was well aware of the demands it would place upon the reader. Not only the typographically trained reader, but also the cognitive-scientifically schooled researcher. I was also well aware that not every reader would be willing to expend the amount of effort that might be required to make sense of it, and I was thrilled to note several did. I felt that what would make the text difficult going had more to do with the fairly large amount of unfamiliar expressions, rather than the oddities of construction and syntax. I hoped that the unfamiliar expressions might be transparent enough to allow a level of understanding and not require too much explanation. At one point Pavel and I suggested the possibility of a glossary of terms. So in writing the text I was taking a calculated risk and consciously engaging in a kind of experiment. I thought that what I lost in accessibility, I could gain in precision. Precision, Hrant, about the fuzzy stuff. Precision, not heady euphony.

The type of expository prose you argue for, William, has its benefits, and were I or anyone else to attempt a full, widely accessible presentation of my understanding of perceptual processing in reading, I (or that person) would need to subject myself to the disciplines Williams, and others like him, outline. I hope you will believe me that I am not unaware of these disciplines. But at the time of their preparation, the text in place, and the presentation on which it was based, was what I could manage. The kind of writing it embodies had its own demands.

(Meanwhile, I wonder if Williams means something different with his term 'nominalization' than my dictionary suggests: converting (another part of speech) into a noun.)

I feel it is misleading to suggest, Hrant, that you and I are much closer than either of us to Kevin. There are common themes in your and mine approaches, and I owe a lot to conversations with you in consolidating my own point of view, but as Kevin has stated in off-forum conversations there is much that he and I can agree on. I hope we can get to testing and exploring those areas of agreement and difference eventually with the goal of a convergence of perspectives between the three of us.

>converting (another part of speech) into a noun.

Yes, he uses this grammatical term in its normal meaning.

>I hope you will believe me that I am not unaware of these disciplines.

I don't believe you, sorry. If you knew and understood Williams you wouldn't pile on the stuff he says characterizes bad writing. --such as "not unaware". Try "aware".

William, i'm astonished by the paternalism of your post. While you may be right

> Precision, not heady euphony.

1) I think it's a precision borne of depth, not accuracy. I might not use the term "precision" here.
2) As I've expressed before (although maybe not in public, at least not explicitly) I think your essay is what it needs to be, especially in the "tonal spectrum" of the whole of Typo13. A topic like this can leverage much spirit from what I've been calling this "existential" style. My complaint above stemmed from Nick's apparent preference of style over content, the admiration of the waves on the surface of the ocean instead of the ecosystem within it.

> there is much that he and I can agree on.

Sure. But these are things I would most probably agree with as well! Things like the limits of familiarity, the importance of empiricism, etc. In the triangle formed by our three views, my vertex is quite close to the line between you and Kevin, and closer to your vertex than his. Exactly what do you and Kevin agree on that I don't? The only thing I can think of is perhaps a belief that we can one day have a total grasp of reading (which I have a philosophical problem with) and by extension that it's possible to design type -optimally readable type- in a sound, formal, complete state of mind. This is in fact the crux of the monotheistic/Roman/Western belief system that I was complaining about above; something I myself used to subscribe to in my teens, but no longer.

BTW, I'm not trying to pit you two against each other; I'm all for the covergence -or at least mutual illumination- that you describe; but I can't pretend the triangle is equilateral.

--

> If you knew and understood Williams ....

There's a huge difference between understanding and accepting; it's often hard to do the former, but the latter is a matter of integrity, not ability. I think you're being entirely too dogmatic, even theological.

hhp

What follows is the result of my attempt to translate the formal, precise style of Peter's 8:26 post into the sort of informal, more conversational language more typical of an online forum posting. I apologize in advance to any changes in meaning. Anyway:

I don't want to talk about my writing style so much that gets in the
way of what I'm trying to say, but I think sometimes there's a need
for formal, precise writing. When I wrote my article I knew it
wouldn't be easy reading for most typographers (and even cognitive
scientists), and that a lot of people would be turned off by the fancy
style. Happily, several people did manage to slog through it. I think
the technical jargon was a bigger problem than the way I put sentences
together. I hoped that people would be able to understand the jargon
from context, but even so Pavel and I did think about doing a glossary
explaining the unfamiliar terms. I took a chance in writing so
formally, and we'll see whether that was a good idea. In any case, one
reason I write so formally is that it's easier to be precise in
talking about fuzzy stuff. Not just for the sound of it, Hrant.

William asked for clear, plain writing, and that's a good goal, and if
I were trying to communicate to more people I would hunker down and
make my writing simpler. Believe me, I know how to write clearly, but
I find it difficult. The article I wrote was the best I could do at
the time.

(By the way, I have a hunch that the "nominalization" of the Williams
book is a different animal than the usual linguistics meaning: making
other kinds of words into nouns)

I don't agree, Hrant, that you and I are closer together than we are
to Kevin. We share some themes, and I really appreciate the
conversations we've had, but there's also a lot that Kevin and I agree
on. I hope we can keep testing and exploring the ideas, ideally so
that all three of us wind up on the same page.

I confess, I come from a scientific, empirical background. My mother is a research scientist (although before that, she was an artist. I picked up much of my early fascination with lettering and ornament from her) and my father is a lawyer. One of my best friends is a postgraduate researcher in molecular biology, and another is completing a PhD in (wait for it) artificial intelligence and visual recognition. I myself studied computer science in school, and have spent the last ten years working as a software engineer (which is something very different). It's possible I have a prejudice towards falsifiability and materialistic analysis of data rooted in quantifiable phenomena (although I'm not sure where "monotheism" is supposed to enter the picture).

At the same time, I'm a reader first and foremost. Thanks to an early exposure to critical theory, I'm comfortable with "difficult discourse", and I've come to see the value of inductive, intuitive reasoning as used by postmodern philosophers like Jacques Derrida and Julia Kristeva. Any philosophy that restricts itself to the world of empirical data is not going to be a very useful philosophy, and there are few experiences more rewarding than the "aha" that comes with reasoning my way through a knotty bit of metaphysics and finally grasping it for myself.

Reading is both a physical process and an act of cognition. One is the domain of research, and the other of cognitive psychology and philosophy. I think that if we're going to talk about the readability of the *form* of a text (that is, its layout, typeface, and other formatting choices, independent of the words composing the text itself), we have to unbundle the overloaded word "reading". There is type, there is layout, and there is text. We recognize letter- and / or word-forms, we follow their flow across the page, and we process the text itself. These are all independent acts, some more mechanical than others.

Hrant, you clearly imply in your article that you value the notion of "readability" over "legibility". You seem to believe that research into legibility that fails to account for readability is compromised or even useless, but I'm unable to determine why you believe that readability subsumes legibility in such an overwhelming way. It seems to me that the research Kevin Larson summarizes (not synthesizes; his paper is clearly an effort to help typographers understand the state of play in cognitive research, rather than to advance any thesis of his own) is concerned with the mechanical aspects of reading and how they relate to legibility, and I don't really see where or how you undercut that research, except to dismiss legibility as unimportant next to readability. When you do try to attack things head-on, that's exactly where things start to get woolly.

For example, you say that undergraduate students may have less reading "ability" than post-graduate students, but I don't see this is relevant to the discussion of *how* people read. I have no doubt that I have the capacity to read much faster than I did ten years ago, but I doubt (although am unsure) that the mechanisms I use to pick words off the page have changed. The process by which I analyze those words has changed tremendously, but the mechanical act of reading is essentially the same for me now as it was 20 years ago, when I was starting high school.

You also claim to demonstrate that the PL model of word recognition is inadequate, but only by going far beyond the remit of the studies on which the model is based. What this suggests to me is that we need more data, so we can see if your intuitively-devised theories and the data we can gather will match. I don't think it's enough to say, "I'm sure the way I'm doing things is an improvement, because that's what my theory tells me." At least, that doesn't convince me that you're not selling me snake-oil or just plain deluded.

I keep chiming in on this thread because I want to know how type works. It's easy to say, for example, that serifs work by creating rails of horizontal stress in a text, but I feel uneasy taking that notion on faith. Typography is one of the last great guild professions; there are clearly masters, and yeomen, and apprentices, and there's a huge body of accumulated wisdom secreted away here and there in books. At least some of the conventional wisdom in any profession as old as typography and bookmaking is guaranteed to be false

>the intolerance in which you express your opinion which is only an opinion

Guillaume, my comment was not intolerant, but critical. The tone of it is dismissive of Peter's defense of his writing style. The tone should have been more gentle, but the rejection is thoroughly merited.

Writers, in my view, just as typographers, have a responsibility to their readers. The manner in which Peter wrote is just a model of the kind of turgid, prolix writing that Williams sets out to correct. Further, I find it ironic that in an essay on readibility Peter writes in a manner that needlessly requires the reader to re-read sentences many times just to puzzle out their meaning.

When I write on difficult topics I try to write as clearly as I can. Peter's defense that this style in this essay is more "precise" is unsound. Turgid prose is just harder to read, not more precise.

Where Williams writes of useless nominalization, Fowler, in his "Modern English Usage" writes of "abstractitis" --which is perhaps the best term for what Peter's essay has a terminal case of. Here is an example from Fowler: "'Participation by the men in the control of the industry is non-existent' instead of 'The men have no part in the control of industry.'"

The first edition of Fowler was in 1926, and Williams continues in this tradition with a deeper analysis of the same issues, and more help for those writers who want to correct such errors and make their writing more readable. My harsh rejection of Peter's writing style in the essay is not simply 'my opinion,' but the reflection of a widespread campaign for clear writing that has gone on for over seventy-five years, and has involved the life work of numerous researchers, writers and teachers.

I know very well that some Continental philosophers write with full blown 'abstractitis', and that it impresses many people as deep. Even Nick, a writer of admirably lucid prose, was impressed.

But in reality 'abstractitis' has no defense, except that the writer, regretably, had no time to make it better. The defense that this style, or should I say blunder, is more precise or deep is unsound, and is shown to be unsound in detail and in depth by such people as Fowler and Williams.

Before Forrest started this particular Typophile Forum, Kevin, Hrant and I discussed the possibility of a Typophile Forum on Typo 13. I was eager to see this happen, and one of the reasons was that I was eager to hear how William might respond to my contentions. I had come to understand that William had a philosophy of science background, was familiar with Hrant's contributions to the subject and had read Kevin's paper. I felt my contribution could move the discussion forward. I hoped that William could help sort through some of the misfiring in communication that the subject matter of the issue seems to attract and that he could help keep attention focused on what needs to happen to resolve the differences in interpretation.

In my Typo text I tried to articulate my perceptions as clearly as I could, given the Hrant (bouma), Kevin (parallel letter recognition model of word recognition), varied empirical studies context and I resorted to continental-philosophy-style rhetorical devices to do this. I don't fault Parmenides, Martin Heidegger and others like them for bending the resources of the language to convey what they believed needed to be brought across. They did what they thought needed doing. Reading them with the expectation of learning from them, or gauging the merit of what they have to contribute, implies a willingness to give them the benefit of the doubt, also in matters of expository style.
To the extent that the writer of a text has a responsibility toward the reader, the reader, who wishes to make sense of a particular text, has a responsibility toward the text. And that responsibility is to take the text on its own terms and become equal to the challenges it places before the him or her. Not everyone will want to make this effort. Many will rely on the judgment of others. If the reader thinks the same thing can be said in more expositorially 'correct' ways, without loss of information and reach, than he or she should make the effort.
I am not averse to pursuing such a challenge, or at least exploring, with others, the possibilities of restatement. But I do hope William will feel inclined to address specific claims.

> I do hope William will feel inclined to address specific claims.

I was completely put off by your writing to the point that I did not want to read it although I am quite interested in the subject matter and read (but didn't study) Kevin and Hrant's contributions. But I am now working through your essay, and will post when I have finished - which may take a while.

Peter, I am sorry for the intemperate tone of my post last evening, which as Guillaume pointed out detracts from my argument. The fact is, though, that Williams' work is rooted in linguistics, and is supported, I believe, by contemporary research in cognitive science.

For what it's worth, my view is that clarity or lack of clarity in writing is normally a much bigger factor in readability than is typeface and layout. Jenson and Griffo I think figured out how to make type more readable than the scribe's pen, and type ever since has been generally excellent in readability, with a few unfortunate detours.

William, now, i'm not astonished anymore, but i admire your confidence. That way you have to carry the truth and expose it to the poor ignorants that are enough dumb to have other opinions than your masters, is brilliant.

I mean, the fact a scientist you like, and others, claim something doesn't make it blessed truth. It makes it an opinion, informed and intelligent, argumented and build, but still an opinion. Let's say a thesis of you want. But i'm more than sure i can bring the same amount of scientists to say the contrary with admirable deepness.
More: the exemples you give are irrelevant, because in each case they use a easy sentence, while what Peter speak about are complicated things. I don't even want to argue about that precise topic, but there are tons of arguments to defend a difficult writting, if it's used to carry a complicated meaning.
Anyway, what disturb me isn't the thing itself, it's the arrogant way you don't even think it's worth talking about it.

>poor ignorants that are enough dumb to have other opinions than your masters

Guillaume, I get the feeling that whenever I contadict the views of your masters you cry 'arrogance'!

I have never met Williams, and he is only my master in that I have read his book and try to follow it. I do know that there is a long effort of research, writing and teaching on clear writing. I think that the arrogance is not on my part in recognizing their contribution and making an effort to learn from it.

I am extremely critical of those who ignore what seems to me well estabilished knowledge, confirmed by contemporary research, on what makes for good writing. Not least because it disrespects the reader and smacks of trying to do a 'snow job' on him or her. To me that is an arrogant style.

The examples that I quoted from Williams and Fowler are simple in order to communicate the principles clearly. If you stack up 'useless nominalizations,' unclear technical terms, useless qualifiers, double negatives, passive voice, run-on sentences, then you get needlessly difficult writing, such as in Peter's essay. Kevin and Hrant were writing about the same complex matters, but with clearer writing. (The quality of their arguments is another question.)

In my view, the obscurity of Parmenides and Heidigger are signs of problems in their ideas, not because what they have to say has to be said that way. But this is another story. Here we are in my opinion dealing not with philosophy, but with testable scientific research. State the theory clearly enough to test it, and then do the tests. That will make for progress.

From what i've read, William, your thesis isn't so "well estabilished knowledge, [and] confirmed by contemporary research". Many scientists write in a complicated way, for a simple reason : their audience know the words they use, and then there are no reasons not to use them (words, or specific expressions). Would you expect a doctor not to use the terms he learned when he speaks to peers but explain in detail and with everyone-understandable language sentences ? That's nonsense.
Then, there's indeed a fancy to speak complicated, that i've used myself, and that i've learned to reject as i'm going to be a journalist. But i know in some situation, and precisely experts ones, it's more useful than critiquable.

And William, i respect your (the other William) view on that. It's just the absolute way you express it that disturb me.

Guillaume, you misunderstand me. I am not speaking of or objecting to using clear technical terms to a readership that knows them. Neither I nor Williams and Fowler have any objection to this.

If you look at either of these authors you will see the kind of problems they diagnose and correct.

If you are going to be a journalist, I think you will find Williams helpful, though it depends on the sort of journalism. Williams is concerned with helping those who have to write as part of their work, doing reports, etc, on difficult matters. So the more complex the matters you have to write about, the more helpful Williams will be. (I think his stuff is almost all equally applicable to French language.)

Generally speaking I find that journalists write better than everyone else, perhaps because they have a lot of practice. The usual problem with journalism is not bad writing, but lack of depth.

Raph, that's cool. But could you make me sound like Peter? :-)

Forrest, I now feel much more comfortable with you - thanks for writing all that up. And I for one don't think your background is a problem here; I'm a CompSci guy too, after all. I think it brings more to the table than takes away.

> we have to unbundle the overloaded word "reading"

Yes, we have to see its dimensions (although I think context still allows us to simply use the word "readability" for the different aspects). In this context of Typophile, the word/concept refers to the micro typeface level. Sure, not world-altering, and yes, less important than the layout, and certainly the language, but that's what we've chosen to be here for (as individuals), so that's the focus*. This is a healthy sort of separation - which however doesn't mean it should become an end to itself.

* William, if we're really concerned about people (like those whose houses are bulldozed), we would worry about much bigger things than sentence structure... But these are our circumstances, and choices.

Readability versus legibility: It's not that I think the former is a superset of the latter, or that research effort must always address both, it's that readability is more complex, more misunderstood, harder to optimize, and most of all more useful to optimize. Letters have to be extremely strange for a person to give up reading a poster, but even a slight -consciously undetectable- readability defect in a text font can impede completion of a book, or even an article (especially if it's something you have to read, as opposed to enjoy). That said, and as Peter focuses nicely on, the essential decipherment that goes on (at the letter and bouma level) is very important to understand, and that applies to both. It's just that I'm personally in the dark on that - as is virtually everybody else though, it seems.

> you say that undergraduate students may have
> less reading "ability" than post-graduate
> students, but I don't see this is relevant
> to the discussion of *how* people read.

This refers to a classical issue: the different types and degree of reading - which can get quite confusing. The relevance here is this: every [adult] reader uses boumas; but the degree to which multi-letter boumas (and the parafovea) are relied upon depends on experience (among other things); and the key here is that empiricism has [had] trouble detecting this. This is where the difference between what might be -awkwardly- called light versus deep immersion comes in: when a test subject relies very lightly on multi-letter* boumas, the people analyzing the data will simply not see what's happening. When they see something like that "and" being skipped for example, they sort of improvise a tentative explanation, and say more research (ie money :-) is needed; their "scientific integrity" (at least the contemporary fashion of such, not the old Greek style) prevents them from doing what I've done.

* BTW, is there a nice tight term for "more than one"?

> we need more data

Agreed. And we need better data, and better interpretation of it all, by relying on "practitioner review" more than peer review, by not staying aloof towards anecdotalism.

> "I'm sure the way I'm doing things is an improvement

I'm never sure. That would be too Modernist. :-) I only think, and try to act on it. I don't believe in the sagacity of the PL model, so I put serifs on my text fonts for example. I could be putting serifs because that's what the status quo wants, but we already know what I think of the status quo... I do it because it makes [functional] sense. And in your case, if you're really convinced of the PL model (your other choice is to be closer to my view - I can't think of an alternative - although there might be one, or even many) then you should only use serifs fonts for aesthetic effect.

> It's easy to say, for example, that serifs work
> by creating rails of horizontal stress in a text

And if you did you'd be joining in a naive delusion. Leading is what keeps lines together (assuming "normal" word and letter spacing). There are no rails. And we don't need arrowheads on the right sides of serifs either. Serifs bind. They reduce letter individuality in favor of bouma integrity. This btw is where legibility and readability are opposed, although of course not fully opposed. Think of them as two vectors at 90 degrees.

> The reason for the hostility towards anecdotal
> evidence in science is because, in effect, it
> requires us think theologically

I think it's because it requires us to think, period!

--

> a widespread campaign for clear writing that
> has gone on for over seventy-five years

Is it a coincidence that this same time period has seen the darkest hours of humanity?

> The usual problem with journalism is
> not bad writing, but lack of depth.

And you think this and their alleged "good writing" is another coincidence?

> well estabilished knowledge

YOUR chosen knowledge, not everybody else's.

--

> I was eager to hear how William
> might respond to my contentions.

I can think of only two people who have the required depth, breadth, objectivity and intellect to serve as a "moderator" if you will concerning the contents and intentions of Typo13: Herbert Spencer (dead), and Richard Southall (retired?).

hhp

For those who are interested, here and here are brief summaries of Williams' book on good writing.

My friend who is director of the cognitive science program at Georgia Tech tells me that the 'storytelling' model of human understanding is now scientifically recognized as one of the best. This confirms Williams' view that when writing follows that model, it is most direct and easily understandable. When it departs further and further it becomes harder and harder to understand.

Williams has also drawn upon much research in linguistics, as he acknowledges in his book. Of course Williams' book is much better than the summary, but this will give you a taste of what I think is an outstanding work.

William,

Sometimes, complex subjects require complex statements. As far as I'm concerned, my first responsibility when I write for publication is to myself - to write precisely what I mean. If I can then make what I have to say accessible to a non-specialist audience, then I will, but its not a big priority. I have this luxury because hardly anyone but a specialist in quite a narrow field would ever read what I've written. Peter is also writing for a specialist field. This article is only likely to be read by people who want to understand technical issues surrounding readability & legibility. If people want to understand those issues, then they are going to have to put some effort in and concentrate. I realise that you're saying that clearer writing will allow readers to concentrate more on the arguments, but I really think that you're too dogmatic in berating Peter for his writing style.

I haven't read the book by Williams that you refer to, but I do use Fowler sometimes. Lets not get carried away with it, though, as Fowler was a pedantic stuffed-shirt and an appalling snob.

I have not yet read the TYPO 13 articles, so I'm not going pontificate as I usually do. But I am intrigued by the possibility that empirical research might answer some of the most difficult questions facing typographers, in particular how to best space type. I really admire the work that the Clearview people did in testing different letterspacings to determine the optimum legibility. Perhaps analogous research can be done for readability.

The questions I would most like to see answered are:

1. What letterspacing optimizes readability? In Jenson's 1470 Eusebius, the space between nn is almost the same width as the inner space of the n. By the mid-16th Century Comensis Episcopi Nucerni (I'm working from a scan from Updike's <i>Printing Types</i>), that width had become about 3:4. And of course 1960's and 70's phototype took the trend to an extreme. Who's right? To my eyes, the Comensis is a striking example of beautifully spaced old metal type.

2. Should the goal be a completely even page color, where all spaces between letters appear to have exactly the same width, or do subtle variations in spacing help with word recognition?

3. Assuming that subtle variations are desirable, should they be based on metal type standards, or some other principle? To my eyes, pe and oo are two common round-round pairs that usually appear tight, while ob and oy are usually loose. These variations arise naturally from metal type technology, and have survived into the digital age.

4. Should letter spaces be absolutely consistent, so that all images of the same letter-pair look identical? One thing I see frequently in older printed samples but very rarely in modern works is a hairline space slipped into a word. I have not discovered the logic behind the placement of such spaces (Burnhill touches on this point in his book Type Spaces), but am very curious to find what effect they had on readability.

5. How does size affect optimal letterspacing? It is nearly universal for smaller sizes of metal type to be spaced more loosely than larger sizes, but this component of optical scaling is usually missing from digital typography.

Of course, the interaction between letter shapes and spacing opens many cans of worms, because of the huge diversity of letter shapes. I'd be interested in answers to the above questions even assuming the letter shapes are from a traditional roman text font.

Great questions, Raph.

Steve, whatever his personal shortcomings - and I have no idea - Fowler did have a great eye for some things - like 'abstractitis' - that make writing worse and better. The fact that it is still in print 75 years later is a tribute to the fact that it was a breakthrough book in its field.

Williams is better as his advice is more flexible and nuanced. For example, he is not simply for or against nominalization, for or against passive voice. He tells you when each helps and hurts. I suspect that his advantage is that he applies linguistic insights developed since Fowler's day. He also thinks issues of correct word usage - the usual obsession, including Fowler - are of secondary importance.

Steve and Guillaume. On complexity, I think writing should be as Einstein said of scientific theories: as simple as possible, and no simpler. What I reject is needless complexity that just makes the writer difficult to understand. I'm afraid that Peter's article has much of this, but I am reading it for content anyway.

I think those are terrific questions Ralph.

One critique on my TYPO 13 contribution is that it does not provide enough detail to address any such questions. This is a valid critique. Unfortunately the state of the art in the psychology of reading is not to the point where understand reading nearly well enough. Of course progress is being made, and there are many terrific people currently working on the problem.

In my opinion the biggest omission in my paper is any discussion of letter recognition. The crux of the paper is that letters are the primary perceptual unit in reading. The paper presents evidence that we are using letters to recognize words both in the fovea and in the parafovea (contrary to earlier statements I do believe we use the parafovea to recognize letters as far as 15 letters out from the point of fixation). The next obvious question is how do we recognize letters. I need to answer that question satisfactorily in order to have a practical impact on type design.

Peter

> the intemperate tone of my post

William you say your friend's information confirms Williams' view that "when writing follows [the storytelling] model, it is most direct and easily understandable. When it departs further and further it becomes harder and harder to understand,"
And in the preamble to the material in your first link Kenneth Mahrer quotes Williams as saying: "Do not take what we offer here as draconian rules of composition, but rather as diagnostic principles of interpretation"
I can more easily acknowledge the accuracy--and appreciate the relativity--of 'diagnostic' statements like your "when writing follows" statement than I can absolute judgments with good and bad as their terms of reference.

If a story-telling model of human understanding is 'the best', what does this mean for thetic-critical discourse? Does it mean thetic-critical discourse must be avoided or recast according to a storytelling (character+actions) model? And is this always possible?

Yes, Raph's questions are good questions. I suspect each of the three of us--Kevin, Hrant and I--have a different take on questions of spacing. I'll try to articulate my take in a future post. The seeds of a reply appear near the end of my essay.

And William, time (and further exploration or discussion) will tell if my contribution uses language or articulates a point of view that is needlessly complex.

> I do believe we use the parafovea to recognize letters as far as 15 letters out

Is this something new?

And how can this be, when every single test (at least among the ones I've seen, which are however many) indicates that we can't make out individual letters in the parafovea (especially when the letter has immediate neighbors - think of lateral masking for one thing), even when given many seconds to do so, much less during immersion? I've never seen anything over 5 or so - certainly not fifteen!

> I find his ideas about creating harmony between the black and white of text compelling.

But you do realize of course that he's talking about a harmony among the black and white of a bouma, not [just] letters. Most significantly, he applies this to foveal processing too; he places even less faith in the PL model than I do. As we say in Armenian, let's sit crooked but talk straight. And I'm not trying to start a fight between you two, so please don't accentuate your agreements as a way to build fortifications against me - work with me here, let's spill all the beans.

BTW, I agree with the centrality of this harmony too, of course; notan is what we read - this is not news. One difference though between Peter and me is that he wants to see the insides and outsides of boumas weighted equally no matter where they're seen (fovea or parafovea); this is one big place I think he's being too Modernist. To me, the weighting depends on the distance from fixation (for obvious reasons of blur). I think this appeals to me and not him because it's a messy ideological wrench-in-the-machine. It's filthy human effluvium.

--

The mechanics of bouma decipherment is one thing, the belief that boumas exist is another. The former is where notan is the issue (and it's no small matter for sure), the latter is where cohesion (think serifs) are the issue. To me we're discussing the latter - not least because I think serifs, tight letterspacing, etc. do more to bind than bad notan does to hinder reading; we can get used to new shapes, but we can't learn to bind with no glue. Slap serifs on a "funky" skeleton (as long as the degree of divergence is correct), and space it well, and you'll outperform the most notanically-optimal sans in the world (Legato). Well, I think so at least.

Reeling in a bit:
The PL model says you can set a long book in sans and not affect reading fatigue. There is no empirical evidence to counter this view, so the empiricists are happy. There is however the little matter of five centuries of vehement anecdotal opposition to this idea. How do we explain this disparity? But throwing the entire corpus of anecdotal evidence out the window, or by accepting the limits of the PL model? <woolly>Come on.</woolly>

> I will of course share my findings.

Which is very generous. I've said this before, and I'll say it again: we're very lucky to have you, and I think and hope that your role will be pivotal in the evolution of the understanding of readability among practitioners. But you will have to be patient with us.

--

Raph, [my] answers soon.
But read the stuff, dude.

hhp

>>whatever his personal shortcomings
I was referring to hs professional shortcomings...


>>the 'storytelling' model of human understanding is now scientifically recognized as one of the best
I think that would be hugely dependent upon the content of what one is trying to understand.


>>scientifically recognized
Perhaps it is Peter's reluctance to use entirely valueless phrases such as this that makes his writing too long-winded for your tastes.

> Is this something new?

No. I quote from the paper (talking about the moving window study),

> we are using additional information further out to guide our reading.

Sure, I remember that. But, as I rebutted in my own essay*, you do limit the usefulness of this information to things like gross word boundaries which help determine the subsequent fixation point. This is not at all the same thing as saying we can make out and treat as usable information individual letters that deep.

* "The PL model ... only uses general information from the parafovea".

> average saccade length is 7-9 letters

Yes, average...

> Peter would not restrict the primitive used in perception to a single letter.

As of course neither would I. And to me this is the big issue here, if partly only because the notan-decipherment issue is: even harder to grasp; and I feel much harder to apply towards practical design decisions. If we could simply settle the serif-sans thing for example (which does not [necessarily] get [fully] into notan), that would be huge.

hhp

> but Peter would not restrict the primitive used in perception to a single letter.

That's right Kevin, but I would also argue that the primitive used is different for different tasks. For example, the perceptual processing primitive might be a single letter in letter recognition tasks, or in the 'letterwise decipherment' underlying learning to read, or in the acquisition of unfamiliar words, or even perhaps in the checking up on an initially misread word.

I append 'perceptual processing' to 'primitive' in this contact, because what constitutes a primitive to the perceptual processing involved in reading or the perceptual processing involved in the recognition of letters, globally considered, might not be a primitive relative to neurons in the initial layers of the visual cortex. In fact I think it isn't.

In my last post I said:
> I append 'perceptual processing' to 'primitive' in this contact, because what constitutes a primitive to the perceptual processing involved in reading or the perceptual processing involved in the recognition of letters, globally considered, might not be a primitive relative to neurons in the initial layers of the visual cortex.

I might have said: ...what constitutes a primitive relative to the neuronal complexes in the visual word form area might not be a primitive relative to neurons in the initial layers of the visual cortex.

> people still gayly choose to favor the style of one and the implications of another

Hrant, I don't know why you assume my choices to read or not read were based on style. I stopped reading Kevin's when he gayly dismissed the dominance of upper and lower case over all-caps as an accident. That's like telling a chef that tasty meals have nothing to do with nutrition. If I were a scientist, I would be inclined to create tests that explain the prevalence of U&lc, rather than gloss over it.

> What letter-spacing optimizes readability?

Ralph, your question hits to the core of the issue. Quite plainly, there can be no scientific answer to this, because it is a design problem. Let me explain.

However reading is conceptualized and measured, whether by words, letters, strokes, or other more abstract, less representational visible features such as those mentioned by Peter, the job of the type designer (and typographer) is to create typesetting which is a harmonious image field against which the individual components may be optimally differentiated.

The reason for the harmony is two-fold. Firstly, harmony provides a consistency of "colour" against which meaningful forms stand out clearly; uneven colour creates artefacts which may be misinterpreted as significant. Secondly, when the discrete nature of letters is compromised, the harmony of "all separate letters" is broken: this disharmony occurs when, through type design flaws, accidental ligatures occur, such as "r_n" confusing modern with modem. Similarly, the letter-components of words may configure themselves ambiguously, for instance "seklan" might be read as "seldan".

No amount of theoretical knowledge will help designers solve the problem of "even colour", because the design parameters are too multi-dimensional. Let's face it, even two parameters require a human value-judgement, as in "do you want that (design product) lighter or stronger?"

Here is a demonstration of the value-judgements that have to be applied to spacing. The test-case is the spacing (and possible kerning) of the sequence "OTC" in a sans serif italic. I have chosen Italic, because it's trickier, and highlights the difficulties.



1. Roman.

2. The problem arises because the O and C should not be straightforward skews. That doesn't look right -- among other things it distorts the thickness of the supposedly monoline stroke, and such inconsistencies (including optical illusions) are compensated for in glyph construction. For the same reason, vertical strokes are thicker than horizontal ones; why? well, that's a whole nother ball of wax, as Peter might say.

3. Because of the compensation, the distance between the glyph near points becomes uneven. But everything must be evenly spaced for optimum differentiation. So how does the type designer resolve the problem? It's a subjective value judgement as to what is most important, which is, in effect, a decision as to what relative priority the reader will attach, in his/her "sense" of even colour, to the different formal relationships involved in this glyph sequence.

Here, in 3, Neue Helvetica 46 ships with no compensation, the presumption being that the distance between the major vertical strokes should be consistent.

In 4, optical kerning (InDesign) attaches more importance to near-point distance, with the result that the larger-scale value of white-spacing becomes uneven.

5. Here, in a spacing strategy more often seen in display setting, the near-point values are equalized, with larger scale gross unevenness; but who is to say that this is not more legible in immersive reading, until it is tested empirically?

So, there are different spacing strategies available, the relative import of which may vary with the size, tracking, weight, and style (eg degree of slant) of the typeface. These strategies have a bearing on glyph design: for instance, the top and bottom extrema points of the O in Helvetica 46 are "skewed" less than the straight "vertical" strokes. By how much may be decided by issues such as this spacing problem.

Furthermore, resolving this spacing problem may upset spacing relationships in other character combinations.

This particular example, concerning a difficulty of print typography, may not seem significant to the matter of letter-recognition, but remember, even colour optimizes readability, and glyph design is predicated on even colour. This example may not seem significant for reading roman text on screen, but it is typical of the complex design considerations involved in type, and not something that can be reduced to simple science.

As Peter says, in boldface, the "micro-aesthetics" of typography are actually functional.

Now, on to the biology. It is my understanding that there are different receptors, or processors, in different areas of the visual cortex, for things such as "closeness", "verticality", "evenness", etc. These are used in parallel.

Prompted by an initial stimulus, the viewer creates a "bouma" which he/she throws into the rapidly cycling feedback loop of reading. This bouma is a flexible, predictive model of what forms the reader expects to see on further inspection of the text. The reader is searching for a match between linguistic meaning and textual form, both of which resolve in unison. As John Hudson has mentioned in a previous thread, the reader uses whatever is necessary to decode the text. So, at one level, massive tracts can be "quantum read" at speed with paragrpah-size boumas, and at the other end of the scale, the bouma may shrink to sub-glyph size to resolve an ambiguity.

The same text may be re-read: different, additional boumas, deeper meanings.

Boumas of all sizes can be processed by the neural equipment, because it is a bag of instruments not limited by simple representational forms such as letters.

Personal and cultural differences come into play, based on which combination of the visual processors are used to read the text. For instance, one person may read "o" as a line which connects upon itself (ouroboros), while another reads it as a shape symmetrical about two axes.

In conclusion, that's it for now.

I'd like to see a moratorium on the use of the word "bouma" as a term of art until we have some kind of consensus on what it actually means. It feels like a neologism that hides more than it shows right now. Is it the boundaries of the chunk of text we're perceiving right now? (This is what I think Nick is describing it as.) Is it the shape of the word currently under your eye? (This is what I've understood Hrant to construe it as.) I don't think our understanding of word shape and textual recognition is far enough along for us to subsume such a tricky and contentious bundle of concepts beneath jargon.

Peter, when Williams says he rejects 'Draconian' rules, it is because he recognizes that different situations call for different approaches. But he is quite definite, as Mahrer immediately explains, that the story telling model is the clearest way to write. The point is that in different situations this goal is accomplished by different means. Also clarity is not the only goal; coherence, for example, is also very important for the reader to follow the writer's train of thought. Hence different situations call for different solutions and compromises, as he explains at length.

He does, however clearly identify what he regards as turgid or prolix prose that he proposes to remedy. If you compare the 'good' and 'bad' examples that Mahrer quotes from Williams, I'm afraid that many of your sentences in your essay follow the 'bad' example for precisely the reasons he explains. And these sentences can be made much clearer, in exactly the manner explained, without any loss of meaning. In fact, you yourself did this above, in rewriting the sentence I pointed out as turgid and correctable.

>Does it mean thetic-critical discourse must be avoided or recast according to a storytelling (character+actions) model?

Please define 'thetic-critical discourse' for me and I can address your question. I do not know what you mean by this phrase.

Williams does not say that all writing must follow the story model, but only that when we depart from it we will be less clear and immediately understood. We may gain something by departing from it, but we will lose clarity.

My friend mentioned the 'story' model in the context of creating and learning concepts in science. If I understood her correctly, the most basic way memories are stored is as scenarios: I open the door door to my house see the stairs on the left, etc. So in understanding the creative process of building new concepts or in explaining a concept to someone who doesn't know it, it is helpful to identify and use such scenarios. If indeed scenarios are fundamental in human thinking--as my friend reports current research seems to support--it does give further credence to Williams' approach to writing.

I do feel awkward using so many words in this thread on the subject of readable composition, when it is about readable typography. However, I will be happy to follow up on your 'thetic-critical' question, as soon as I understand it.

> the primitive used is different for different tasks.

The way I would use "task", it can be different within a task too. Specifically, a bouma is (virtually always) a single letter* in the fovea (because of sufficient resolution - remembering the high parallelism of reading cognition). This I believe Peter doesn't subscribe to (maybe because it doesn't support a tidy notan view).

* This special case is how the bouma model assimilates the PL model.

hhp

>Is it the boundaries of the chunk of text we're perceiving right now?

Yes, but as map not territory.
I agree with Peter that the PL model is too reprsentational.

That's why I'm suggesting that a bouma is a flexible, heuristic concept created by the reader (not a word or letter shape; it might even be used to read alliteration or capitalization of proper names).

I've also noticed that people's ideas are evolving towards a more flexible "primitive", with Kevin and Peter considering sub-character significance, and Hrant likewise moving down in scale. So why not call our "primitive" a bouma?

But I suggest that the primitive is not the basic building block, but a variable construct. After all, there are such things as single-letter words.

William, you say in your 2:35 post:
> Williams is better as his advice is more flexible and nuanced. For example, he is not simply for or against nominalization, for or against passive voice. He tells you when each helps and hurts.

I am glad to hear this (and several additional things in your 6:26 post). I am not sure though that your more categorical posts have helped me gauge which of my nominalizations, double negatives and statements in the passive voice help or hurt. Perhaps you will say 'most of them'. I am intrigued enough to look him up and judge for myself based on what he says.

Thetic-critical discourse is writing or speech that advances a thesis and explores how and why it differs from relevant competing points of view.

I've been pondering whether to make any contribution to the discussion of the style of Peter's essay. On the one hand, I consider this discussion a distraction, and wonder if Peter, Kevin and Hrant's initial notion to begin a separate Typophile discussion on reading might have been a better idea than beginning it within the context of this thread about Typo 13. On the other hand, it is something that I have thought about from time to time since I first read Peter's essay in Thessaloniki last summer, and it touches on my response to philosophical texts I have read in recent years. So please consider this message an aside from what I consider the main discussion, and I'm only sending it because I think some of the exchange between William and others has become unnecessarily hostile. I think there are ways of thinking about Peter's rhetorical techniques that may encourage charity.

Generally, I agree with William that writing should seek to be lucid, and to convey the writer's ideas as clearly as possible. I don't think this always means that the writing must seek to be 'as simple as possible but no simpler': not all discourse is analogous to a scientific theory or, in philosophic terms, a formal proposition. One can, of course, meditate on any text, but some texts are what I would class as deliberately meditative, i.e. intended to oblige or force the reader to approach the meaning through a process of active engagement with the (often dense) verbal structures of the text. I consider Peter's essay to be in this class of texts that invite or demand meditative reading. William criticises Peter's text as requiring the reader to re-read sentences several times to get at their meaning, but this seems to me to be exactly Peter's aim: he wants the reader to wrestle with the text. To read such a text is not simply to take in its information as expressed, but to work at 'translating' it to arrive at a 'version' of the text that is clear to the individual reader. There are few exercises that demand closer engagement with a text than translation, and it seems to me that Peter wants to force something akin to such engagement. Whether we are willing to play along, is a question for each reader.

Where I agree with William is that I don't think Peter's text needs to be written the way it is written, that the ideas in the essay could have been expressed in a more direct and lucid language. But Peter made a decision and employed a deliberate 'rhetorical strategy' to create a text that demands a particular kind of reading, and I'm willing to enage with his ideas on those terms (and it is a strategic decision, which is why I disagree with Raph's translation 'writing style', which seems to miss something of the intent).

At their best, meditative -- or consciously 'difficult' -- texts can be Sokratic in effect. That is, they force the reader to consider assumptions, to return to the ideas discussed again and again looking for a more precise understanding, and they can inspire unexpected ideas in the reader's own thinking on the subject as a product of a meditative process that differs greatly from the kind of immediate response we typically make to what might be broadly called 'information'.

At their (frequent) worst, such texts are simply unnecessarily complex and deliberately obscurantist: they require a meditative reading, but do not reward meditation. In this respect, they are the antithesis of Sokratic: they are sophist, employing rhetoric to mask falsehood, ignorance or philosophical bankruptcy.

The meta-problem of all such texts is that one needs to engage with them quite deeply, and with a degree of philosophical charity -- there might be something of value in them --, in order to determine whether one is dealing with a thoughtful meditative text or with a piece of sophistic showmanship. I suspect that William would ask 'Why should I bother to go to such trouble? By what right does this author make this demand of me? What good will has he shown to the reader?', and I'm often inclined to ask this question myself when faced with such texts.

I've come to the conclusion that one's willingness to deal with such texts is greatly dependent on one's sympathy for the ideas behind them, and that general interest in the presumed subject matter is not always sufficient (although, clearly, negative reaction to the rhetorical techniques employed may overcome such sympathy, as it seems to do for William). I have very little sympathy, for instance, with the ideas of Jacques Derrida, so have only occasionally bothered to think my way through one of his texts (and have generally emerged less sympathetic than when I started). On the other hand, I have much more willingness to engage with a very similar set of rhetorical techniques, even use of Derridean jargon, in Catherine Pickstock's deconstruction of Derrida's ideas on language and writing, because I'm much more sympathetic to her underlying ideas. [That said, the most striking parts of Pickstock's Beyond Writing are those in which she is most lucid, drops the jargon and writes in plain English. But perhaps that too is a rhetorical strategy.] I'm willing to engage with Peter's ideas in the way in which he has chosen to express them, precisely because I have some sympathy for his critique of both Kevin and Hrant, and I appreciate the subversive nature of his questioning of them.

Like William, when I see a text full of unclear nominalisation, I am usually suspicious of the intent of the author. Often, such rhetorical techniques do seem to signal problems in the underlying ideas, which the writer has attempted to disguise through dense and obscure verbal trickery. I share William's skepticism when it comes to Heidegger, and have never found much of value behind Derrida's verbal cleverness. And I should add that I think the frequent (mis)characterisation of such reactions in terms of an Anglo-American analytical tradition vs. a Continental tradition of philosophy is inadequate, as the oft-remarked-on lucidity of Hans Georg Gadamer's Truth and Method demonstrates. [Derrida's refusal to engage in debate with Gadamer still seems to me a signal admission of philosophical cowardice.]

But I've learned to be charitable, having recognised what I consider an authentic or at least honest strand of such writing, which is subversive of its own technique. Peter's essay, and his comments here, make it clear that the is seeking something through his ideas, and through his approach to 'precision' in their discussion. I'm not suspicious of his motives. I don't think he is trying to sell me philosophical snake oil. I don't think his ideas need to be expressed in the way they are in the essay, but perhaps in the discussion here we can, together, refine and re-express them while maintaining his salutory insistence on precision. As I wrote in the apparently notorious 'graphic communication' thread, I consider conversation to be the very best method for the development of ideas and their understanding.

I also think that there is a delightful irony that an essay that considers 'immersive reading' should be anything but immersive. Knowing Peter, this irony may well be intentional.

Nick: your example is illuminating, but in my opinion speaks more to the difficulty of deciding what constitutes even letterspacing than the

BTW, does anyone have more info on InD's optical kerning feature than what's in this thread? See also this fascinating comp.fonts discussion. I'd be especially interested to know if anyone else has followed up on Chuck Bigelow's ideas of using low spatial frequencies. That idea would seem to resonate rather strongly with more recent theories about the role of the parafovea.

I also found a link to ikern, which claims to be better than InDesign's optical kerning.

John: I see your point about the problem in translating "rhetorical strategies" to "writing style". But what can I say? At the time of their preparation, the text in place, and the presentation on which it was based, was what I could manage.

Re-reading that comp.fonts discussion brought back a lot of memories. I hadn't realised that it was more than a decade ago. Coincidentally, I'm involved in a project now that uses something similar to the sector kerning discussed in that thread (sorry, that's all I can say about it at the moment).

It's hard work to refine language to be concise and clear. As Maugham once said (or words to that effect) of something he wrote, "It would have been shorter if I'd had more time."

John, I think you are quite right about my text and texts like it. But the irony you pointed out at the end had not occurred to me. (I was perhaps too immersed in the rigours of writing.)
Perhaps most of the things that needed saying, pro and con about rhetorical strategy choices have now been said and we can move on.

Forrest, I think a moratorium on 'bouma' might be a good thing. The term is at best a slippery fish. A moratorium might force Hrant to say what he means in other ways.
The use of a technical term in a discussion is a social contract. In the case of 'bouma' I'm not sure every participant in this discussion has agreed to it. Certainly the content Kevin specifies for it in his essay (word shape in the sense of outline or raw pattern of neutral, ascending and descending characters) differs from Hrant's sense of what it involves, or so it seems to me.
In addition, Hrant, the issue as I see it is not 'the belief that boumas exist', but whether the term can be made to embody a construct that is useful for clarifying or elaborating how perceptual processing in reading works. I still believe that might be possible.
Hrant, you say: "...a bouma is (virtually always) a single letter in the fovea..." I think you are wrong about the 'virtually always,' but aside from that, when you say it is a single letter in the fovea I begin to wonder if you aren't playing too fast and loose with the term. Here I begin to think you might be using 'bouma' as a synonym for 'perceptual processing primitive', or 'unit of perceptual processing'.

Hrant's 'bouma' comes from 'Bouma shape' and in that form is first used by Taylor and Taylor in their "The Psychology of Reading." In T&T's text the authors' intent is to factor into a notion of word shape more than 'outline' or 'raw pattern of neutral, ascending and descending characters'. However, in their use of it, a word's 'bouma shape' refers to a coefficient--a string of numbers coding (essentially) perceptual distances between component letter forms. Not 'shape' in the conventional sense. And the point they want to make comes down to this: when this new level of information is added to the notion of word shape, the word shapes of (I think it was) 97 percent of words in a typical text are seen to be unique. So bouma ambiguity in a typical expository text accounts for a relatively small part of the retrograde saccade rate encountered in a competent reader's reading of a typical (Williams approved?) expository text.
Hrant picked the term up from his reading of Paul Saegner's "Space between Words", who adapt it from Taylor and Taylor. In Saegner's glossary its meaning is specified as "the shape of the word when written in upper- or lower-case letters and delimited by space."
In the first Hrant use of it that I encountered there were references to the internal and external outlines in a word. At the time I was reading Gerrit Noordzij's (soon to appear in English) "The stroke," and the reference to the inner and outer form of the word resonated with me. Later Hrant's 'bouma' became the word as seen in blurry parafoveal vision. Bringing boumas into the fovea means that it has to mean something different again than the 'blurry paravoveal vision' iteration.
My own (playful) derivation almost half-a-dozen years ago looked like this: We see a bounded, separated and rhythmically cohering configuration of grouped alphabetic or numerical forms. We see a bounded map = boundedmap = bou[nded]ma[p] = bouma. I might state the first sentence differently now. But the second might still be a place to start. It would allow us to say: the perceptual processing primitive for vision in normal immersive reading of familiar words (where legibility is not an issue) is a bounded map of role-architectural and role-architecturally evoked particulars.
What might be the use to type-involved individuals of saying this, or something like it? Perhaps the following: anything that might hinder the optical integrity of the bounded map should be attended to, whether its source be problems of spacing, glyph shape or stroke conformity. This is why good typographers are fanatical about kerning, ligatures, etc. To my mind we need not think of concerns with spacing etc, first and foremost, as aesthetic concerns, though they can and should be that too. First and foremost such concerns must be concerns to meet an 'ease of perceptual processing' requirement.

This gets us into the 'space craft' thread in this forum. And I will have more to say about that in time. (The Bigelow / spatial frequency grating contribution in the comp.fonts discussion will help!) But also in the picture is the matter of lateral interference.

By the way: how do I get back for reference the top messages on the forum? They have disappeared (because of the 7-day setting).

Ok, Peter, I

> * "The PL model ... only uses general information from the parafovea".

Hrant, I think you are misunderstanding the claims that I made in the paper about the role of the parafovea. The window study showed that reading speed would increase when additional letters are available up to 15 letters. This indicates that readers use information up to 15 letters from the point of fixation to inform reading. The next point in the same section rebutted the idea that readers would benefit from always being able to see to the next word boundary.

I fear there is an assumption that I make about mental processing that I haven

<deleting redundant post>

It might be next Monday or even Tuesday by the time I catch up with everything in this thread, but the following I'd like to quickly address:

> readers use information up to 15 letters from the point of fixation

Well of course. But not individual letter information. And if we can use word boundary information, and partial letter information, maybe we can use more -really, as much as possible- information too, such a letter clusters (that are sufficiently frequent and distinctive, and relative to how clear/blurry they are)? Can't you slowly start seeing the missing piece of the puzzle? It makes everything we think we know and we know we believe fall into place.

> The further away a letter is from the center
> of fixation the less information is gathered.

And the larger boumas get!

--

BTW, the other day I corrected and expanded the "bouma" Wikipedia entry a bit.

hhp

William, all of what you say in your latest post deserves saying, and much of it I can agree with. I am aware of my text's limitations. I was trying to do far too much in thirty minutes and seven large-format pages. Nevertheless I am still, for the most part, proud of it.

Your 'veritable Niagara of neologisms' is apt and amusing. I hope, of course, my Niagara will not come off as a fog machine.

Peter, thank you for your gracious reply on the rhetoric issue, which will hopefully sleep restfully now.

On your article. Here are its two chief claims in a nutshell, as I understand them. Please correct me if I

Nick sees Kevin, Hrant and I 'evolving towards a more flexible primitive' and suggests we think of the primitive as a variable construct.
Meanwhile, in response to my "the primitive used is different for different tasks," Hrant, wants to say 'it can be different *within* a task too.'
A variable construct: definitely. Perhaps we could say the relevant primitive is different for different sub-routines within the different tasks. But the primitive that matters most for the acquired skill of reading an alphabetic separated script, once familiarity with letters is in place, is the visual wordform / multi-letter bounded map.


William, your post appeared as I was formulating the above. You have done a splendid job of presenting my chief claims in a nutshell, and I thank you for putting out the effort. The trouble comes in making such a view defensible in cognitive-scientific terms. And the pressure I felt to begin doing that in the face of a 'hostile' consensus resulted in a thick text. (I hope no one objects to my use of the word 'hostile' here. The prevailing consensus is unfriendly to a position like mine and probably Hrant's, but many of the individuals that hold versions of the prevailing view, like Kevin for instance, are perfectly open to entertaining mine or Hrant's, provided it can be made to seem compelling in cognitive-scientifc terms. I gladly grant them that!)
Your comments about method echo and amplify matters of testing Kevin and I (and perhaps Hrant and Kevin also) discussed when we met in Thessaloniki. Our hope was to pursue this further, and I hope it will come to that.
As to your contribution to this, I for one embrace the notion of a methodological monitor and a grammatical coach, but I promise you, I will not let go of some of my terms easily.

>making such a view defensible in cognitive-scientific terms

Popper would say: Trying to make it defensible in advance may make it more vague and useless. You want it to be consistent with known facts, but beyond that, making it testable should be the goal. Then, right or wrong, you have made a valuable contribution, for new facts will be discovered. Experiment should decide how sound the theory is.

I've now done a first reading of all three papers, and some more digging in the literature. For more on the physiology of vision, I recommend the paper Compound grating discrimination in extrafoveal and amblyopic vision. Also, it's too bad Hrant's blurred fixation images weren't used; they were very helpful for my understanding. I'd really love to see carefully prepared versions of these images based on an accurate model of grating sensitivity and real saccade data.

I can see why people would be turned off by the hranting and hraving in Hrant's article, but I am very sympathetic to his central point, which I will now distort hopelessly by saying in my own words. We can take almost as an article of faith the principle that the visual system integrates all useful information presented to it. Both the word shape and parallel letter recognition models are deeply flawed because they both call for useful information to be discarded.

In particular, I'm talking about the blurred image in the parafovea, and most especially cases where the fixation points of the saccade sequence skip over words (this would correspond to the third of the three zones in Kevin's description, comprising about 8 to 15 letters out from the fixation point). The resolved retinal image does not have enough detail to reliably resolve the letters in these areas, but at the same time it is clear that readers use the image for something. The word shape model would discard all information other than the outline of the word shape, while the parallel letter recognition model would discard almost all information other than the "length of upcoming words and [...] best location for the next fixation point."

I also think Hrant is on to something with his insistence that the reading process is based on heuristics. To continue the "Roadside joggers" example, in the blurred retinal image there is not enough information to reliably distinguish the word "and" from, say, "aml" or "srd". There is, of course, plenty enough information to distinguish it from "from". I'm sure that somewhere inside the visual system there's a representation that looks like "sweat, pain ### angry drivers", where ### represents a fuzzy equivalence class admitting "and" with very high fitness, "aml" and "srd" with high fitness, and "from with very low fitness. The higher linguistic levels no doubt look at this representation and are happy with the fuzzy equivalence class. If they weren't, they'd send word down and there would be a regressive saccade to see what the word really says. I would guess that another function of the regressive saccades is to provide higher-quality training data for the (fuzzy) letter and multiletter recognition discriminators that see only blurry input.

These fuzzy equivalence classes should be familiar to anyone who's tried to do speech recognition. Getting a computer to reliably distinguish spoken "how to recognize speech" from "how to wreck a nice beach" is notoriously difficult. In this example, though, the primary difficulty is distinguishing word boundaries, which is thankfully much easier for printed text thanks to adequate wordspacing.

If my understanding is correct, then the job for typographers is to make images that resolve to as narrow an equivalence class as possible when blurred, so that the linguistic levels can get the same confidence in words with longer saccade jumps. Oddly enough, it would seem that this job is quite similar to that of the Clearview designers, except that in their case the image degredation is caused by fog and night driving conditions rather than the reading process pushing the lower-level visual process to the edge of its envelope to optimize reading speed. So maybe there isn't as much of a rift between legibility and readability as commonly thought.

Nice post Raph! To rich for me to respond to directly now.

William, I did not say 'defensible in advance,' but as you quote, 'defensible in cognitive scientific terms.' This is roughly equivalent with: making it 'consistent with known facts'. 'Making' might be the wrong word. We need to articulate a testable position in dialogue with oestensibly sustainable friendly, and oestensibly sustainable hostile, interpretations of qualified evidence. (Facts are, basically, sustainable interpretations of qualified evidence.) But sometimes resolving a conflict in interpretations is beyond the reach of testing, and other evaluative protocols must be recruited. And this other has to be included in a concept of 'defensible in cognitive-scientific terms'.

Perhaps Hrant's suggestions about single-letter or bi-graph boumas in the fovea and his suggestions about the roles and proportionate contributions of the fovea and parafovea could be tested by turning the boundry study on its head so to speak.
If the parafoveal viewing of word-sized boumas has the capabilities Hrant claims it does, and if the chief role of a foveal viewing of text-matter is to help narrow the equivalence class (using Raph's term) down to one, than a test like the following might prove useful. 1) Create text which uses a word with numerous close equivalence-class or orthographic neighbours. 2) Leave the material eccentric to fixation in place when a saccade is executed, but substitute the letters just around the likely disambiguating fixation candidate of the relevant word with close 'similars'* (or x's )during the saccade (one could play with bouma size here: single letter; bi-graph; tri-graph.) 3) See what happens to comprehension and reading speed compared to a) no substitutions; b) substitutions with near-similars c) substitutions with x's; d) substitutions of all but the disambiguating letter (when the single letter is the most likely bouma candidate under Hrant's scheme) or bi-graph.
The test strikes me as a bit ludicrous. And that is part of the reason I think aspects of Hrant's conception are out of whack.
I predict subjects will find a correct sense of the sense of the sentence might 'lurk', but the experience of reading what is being said will be somehow disturbed.

* see Hrant's "Improving the Tool"

__________________________________

Kevin, you say: "[t]he boundary study described in the paper demonstrates that during normal reading we combine the incomplete parafoveal letter recognition with additional time completing letter recognition in the foveal region.

I think 'demonstrates' is too strong a word here.
Maybe you will grant this readily.
But let me state my reasons for saying it.
Your view of what the boundry study allows you to take as fact might be the most ready-to-hand for you, given 1) the fact that the distal stimulus material is made of letters, 2) your 'parallel letter recognition' bias and 3) your assumptions about mental processes. But it might not be the most probable, elegant, compelling, or fertile, when a wider range of factors and assumptions, or a different bias, or a 'deconstruction' of your terms of reference (sorry about the Derridaian slip John) are taken into account.
Perhaps I need to demonstrate the above, but the most you can say, I think, is: "[t]he boundary study described in the paper is consistent with the view that during normal reading we combine the incomplete parafoveal letter recognition with additional time completing letter recognition in the foveal region." Meanwhile, the most I would grant is: "[t]he boundary study described in the paper is not, on the face of it, inconsistent with the view that during normal reading we combine the incomplete parafoveal letter recognition with additional time completing letter recognition in the foveal region." That is, the boundry study doesn't falsify your view.

The big unknown here is precisely how the parafoveal preview benefit works. Beyond providing length and word-boundary cues, how it works is a dark area. We can be confident of the following: misleading visual cues 1) up into the role-architectural and role architecturally evoked level (bowls / stems / counters and their positional relations), and 2) to a distance of minimally 15 characters from the point of fixation, detrimentally affect the operative neuronal representation of parafoveally previewable material.
I think the effect of lateral interference in a parafoveal presentation of a word is such that Kevin's partial letter information can't be meaningfully resolved and fed forward. Some studies try to work with the notion that a crude but useful neuronally compiled textural map accounts for the parafoveal preview benefit.

I think the effect of lateral interference in a parafoveal presentation of a word is such that Kevin's partial letter information can't be meaningfully resolved and fed forward.

I suspect this is right. I've spent time trying to maintain fixation and develop a conscious impression of what the parafovea looks like (this is harder than it sounds: when presented with text, the brain keeps wanting the eye to jump!) One thing I've noticed is that the loss of focus in the parafovea does not appear to be linearly progressive as in Hrant's blurred illustrations, but then it struck me that this may be due to lateral masking making some letter sequences more difficult to see than others, giving the impression of areas of greater or lesser loss of focus within the parafovea. I also don't discount the possibility that this experience of the parafovea is not primarily optical at all, but that what register as greater or lesser areas of focus are actually the result of my brain being more or less sure of the identity of individual letters in the parafovea. Now I have a headache and my eyes hurt

So, to recap:

We read by recognizing the way words are spelled.

Words with blurry letters that we kind of see, we kind of recognize.

And if you've read most of a sentence, you can guess the

>We read by recognizing the way words are spelled
It seems that Kevin agrees and Peter disagrees. According to Peter we don't mentally assemble letters, but rather identify the whole word as a pattern. I don't see how this part of his views can be the whole story, because of the ease of reading scrambled words. I don't see how you can explain this unless there is an assembly process.

A good place to start in thinking about the 'jumbled word effect' is here: http://www.cs.umd.edu/~shankar/cwhitney/Papers/TICS.pdf
You might also want to check this out (it's more recent and I haven't done more than skim it): http://www.up.univ-mrs.fr/wlpc/pagesperso/grainger/pubpdf/Carreiras&Grainger04.pdf
A long version of Whitney, et. al.'s coding scheme is here: http://www.cs.umd.edu/~shankar/cwhitney/Papers/PBR.pdf

Before going too far in providing an explanation I'd want to see how scrambling inner letters affects reading speed, and I'd want to see a saccade / fixation / retrograde saccade plot.

William is right: The visual wordform resolution part of my story is not the whole story. It applies to orthographically regular versions of words for which visual wordform pathways have been consolidated.
In the case of jumbled words, I'd explore the possibility that, when a jumbled word is first encountered, a letterwise (or familiar bi-gram / tri-gram wise) assembly operation kicks into gear. That is, a perceptual processing routine analogous to those we use when learning to read or encounter an unfamiliar word kicks into gear. Perhaps once the reader's 'readingware' is on to the conceit, the assembly operation becomes more automatic.

I

Hrant,
I

Saying that images are blurred in the parafovea may be a bit of a shortcut. Probably a more precise statement is that sharp and blurred images on the parafovea are perceived more or less the same. This paper gives a fairly striking demonstration of this effect.

What really happens in the parafovea still seems pretty mysterious to me. The compound grating paper I linked above strongly suggests that you lose the ability to perceive phase information more quickly than you lose the ability to perceive contrast information as you go from the center of the fovea. That's a fascinating nugget of information, and one that I'm pretty sure is relevant to typography, but on the other hand, it also shows that you can't just do a Photoshop radial blur and expect it to accurately simulate the perception of images in the parafovea.

>Word recognition is mediated by phonology, or the sound that a word makes.

This proposition is common sensical, like the idea that we read by spelling words.

Are there any relevant reading tests of deaf/dumb people?

One way that the visual differs from the spoken is in the amount of words available for conscious perception; consciousness has a 4-second "moving window" duration (sorry, I can't remember the research details), and in that span the field of vision can contain many times more raw text than spoken words (discounting polyphony).

What of saccades that jump from paragraph to paragraph? Written text is frequently created with this expectation. Perhaps it is a principle of information structure that certain bureaucratic and legal documents are arranged by hierarchical, number-tagged paragraphs, but they are also read that way at first encounter, in light scan mode. Same thing with the PowerPoint bulleted presentation style. And lists like timetables.

Imagine you are skimming the boldface names in a gossip column: how do you read those? I would assume that you blow right by celebrities you don't know or rate, and fixate on your faves. You wouldn't read the whole name (like those people in Russian novels always named in full with long unpronouncable names, whom the reader mentally refers to as Count V______ or A_____ A_____). So you'd be primed to recognize names like Brad or Gwynneth. Why would anyone read the whole word or mentally sound it? Surely the input "four letter word, begins with B" requires no further recognition -- until you're halfway through the sentence and discover you're reading about some playwright called Bard.

But this is not immersive reading, it will be said. To which I reply, define immersion. And what's so immersive about lab tests that measure Word Recognition of separate- character type in terms of speed?

Is reading active or passive? Certainly when you measure it in a lab, it is passive. As Kevin says, when forced, people can read all caps as fast as U&lc. Research funded by Microsoft.

Alternatively, when reading is active, the reader defines the terms of immersion, the kind of information he/she wants to get out of the text.

The assumption is that word recognition tests can establish some basic, universal principles of reading, which, some believe, will help typographers and type designers. But these will not be universal, because they are based on passive reading of roman type (not handwriting), and the results will be crude and nasty.

Well, those with power get to define reality. I would expect that in the future, generative typographic technologies will be developed from the understanding of reading acquired in the world of corporate research. These will be based on reductive models, driving the human interest into more complex, organic forms -- handwriting in particular -- and into more advanced theroretical areas, such as quantum mechanics.

The reductive, first generation artefacts will be refined, as in the way "we" have moved from Arial to Calibri. But in the meantime, progress means an awful lot of Arial.

About what really happens in the parafovea, the following might be of use:
http://visionlab.harvard.edu/Members/Patrick/PDF.files/2001pdfs/seeing.pdf
and this is tanatalizing:
http://journalofvision.org/3/9/659/

(the actual papers cost, and I haven't shelled out)

Kevin, is the time data that motivates your 'funny claim' from lexical decision tests or fixation-time readings (in sustained reading of continuous texts by competent readers)? If they are from the former I suspect they are less relevant than you might think, to issues of perceptual processing in sustained reading of continuous text using competent readers. If they are the latter, I will need to see if my model as it stands can accommodate them.

I do not doubt that what you say is relevant to real or laboratory-simulated lexical decision tasks; and I also hold that lexical decision is relevant in continuous reading when the automatic processing that normally subtends it is interrupted, but I do not hold that each visual wordform resolutional event involves a lexical decisional act. I propose that a saccade is executed when visual wordform resolution happens, not when a lexical decision is made. A sense of the unfolding of sense while reading (and a sense of connectedness with what is being said) happens--and this is the real mystery--without the aid of multiple discrete lexical-decisional mental acts, or so I think.

I hold visual wordform resolutional events are the crux of learning to read, and the engine underlying continuous reading of larger texts.

Raph, the representation of parafovea and peripheral vision in the paper you link to strikes me as correct. That

Just to follow up on testing issues. I see from your post, Kevin, that testing is being done on Chinese. I don't whether they are asking questions about spacing (within characters), though.

Incidentally, I am aware that there is a phonetic element in Chinese characters. There are several hundred 'radicals' if I remember right that are used to construct other characters. Typically, they will combine a character for sound with one for meaning, giving a clue to the reader who knows the system. I am told that after the first few thousand characters it gets easier. I never got that far.

On the Hebrew vowels, I think there is something to be learned here, as they are pretty disfunctional, I suspect. Hebrew is normally written without the vowel points (nikud), both traditionally and today. The vowel points were introduced I believe to preserve the pronunciation for a populace that was losing Hebrew. I suspect that even native speakers cannot read Hebrew with nikkud 'immersively'. In other words, they will just ignore the vowel points. -Some Hebrew speaker can confirm this for us. And I don't think those who learn to just pronounce Hebrew phonetically (for prayers, without understanding) ever get to the stage of words 'gelling' so they just see words. If this is the case, it might tell you, negatively, about what interferes with immersive reading, and hence some clues about what is good.

Here is what I think we could work toward in this forum:
1) resolution of the key issues / a convergence of perspectives;
2) a broadly accessible articulation of the resulting perspective;
3) exploration of impacts for font generation / typography
Are there opportunities for a convergence of perspectives?

Building on elements from the discussion so far and bringing in some new terms, can a scenario such as this be of use:
Let us say 'n' is a fixated word, and 'n+1' is a subsequent word in a line of text accessible to parafoveal view.
1) Using 'ensemble statistics'* drawn from parafoveal pre-processing of n+1 material, magnocellular processing mechanisms** within the visual cortex narrow the visual wordform candidate field. The narrowing process sets up an activation candidacy set.
2) A saccade to a criterial area inside the bouma, or word, narrows the field to one and sparks a visual wordform resolutional event. Or to say the same thing differently, a representative sampling of local statistics will close the visual wordform resolutional loop.
3) We connect to the distal (space-delimited) stimulus complex--the word. It appears to us as object-like, familiar, sense-invested, and sharply defined or fully articulated inside and out (even beyond the normal 'range of high acuity' span).

The interesting testing question might be, how narrow or broad is the typical fuzzy equivalence class parafoveal previewing brings forward.

* see: "Seeing the forest but not the trees"
http://visionlab.harvard.edu/Members/Patrick/PDF.files/2001pdfs/seeing.pdf

** see: "Microgenesis, immediate experience and visual processes in reading" by Victor Rosenthal
http://cogprints.org/2489/01/MicroRead.pdf
[page 10 &11]

ad 1): A) We don't know precisely what the ensemble statistics are. Presumably the members of a fuzzy equivalency set have identical or comparable ensemble statistics. Changing one iota--changing any role-architectural particular--alters, we must assume, the ensemble statistics, reconfiguring, however slightly, the fuzzy equivalency set. This could account for the boundary study results. B) Magnocellular processes narrow the field by by alerting neural pathways joining members of a fuzzy equivalency class. C) In the fuzzy equivalency set are objects whose phonological, orthographic and sense*** profiles differ. Call them P O & S profiles for short. These factor into saccade planning. But in continuous reading of familiar words they do so by proxy and without representation. Part of learning to read is learning where to land--where, given a particular set of ensemble statistics / fuzzy equivalence set / candidacy set + their attendant P O & S profiles, to touch down. Saccade-target learning is an intelligent heuristically-driven enterprise. Because P O & S profiles play a role in learning the optimal viewing positions for each fuzzy equivalence class,leading to expectations of what local processing might find phonological priming tests show a facilitory or inhibiting effect.

*** has to do with locations in personal construct space (near points / far points in personal construct space)

ad 2): Because only a representative sampling is required to close the visual wordform resolutional loop it is attractive (but mistaken) to posit, as Hrant thinks we must, that the bouma is perhaps only a character or two long in the fovea. Mistaken, because the relevant primitive to which the perceptual system's response bias is skewed remains the visual wordform as a whole (as indexed in parafoveal preview).

ad 3): Visual wordform resolution is an individuating event. It mediates directly the unfolding of sense, because, encapsulated with the visual wordform's neural signature is a location in personal construct space. (This is a topography and not a database) The unfolding of sense does not involve multiple discrete lexical-identity assignment events.

In short: parafoveal processing: responsible for 'index assignment,' 'ensemble statistics assembly,' and 'fuzzy equivalency class selection', or 'candidacy set narrowing'****; foveal processing: responsible for 'individuation' and triggering the 'visual wordform resolutional' event.

**** for objects for which neural pathways exist. (for: 'index assignment' see "Indexing and the object concept: developing 'what' and 'where' systems." Alan M. Leslie, Fei Xu, Patrice D. Tremoulet and Brian J. Scholl: http://ruccs.rutgers.edu/~aleslie/Leslie_%20Xu_Tremoulet_&_Scholl_1998w.pdf)

I’m not sure if I’m doing right in contributing to this trhead with my post but I think that altough I’m no expert on the topic, I can provide a view that may help to elucidate the problem. This is the view of Frank Smith’s “A Psycholinguistic Analisys of Reading and Learning to Read” (I acutally read the Spanish translation).

I read the three articles this afternoon as part of the research I’m doing for my own thesis and so I decided to jump in.

Frank starts stating that the reading hability is not different from mundane perceptual activities such as discerning tables from chairs and that there is nothing in reading (speaking of language) out of the competence of anyone who has the hability to understang spoken language.

His main statement is that reading is not about deciphering or extracting sounds or words out of the text but about giving it a sense (reading is a perceptual activity and a function of perception is to give sense to the information our senses report to the brain and to guide them in the sake of such).

He distinguishes two kinds of information used during the reading process:

* Visual Information, which is the images the brain recieves from the printed matter. Visual information is indispensable but not enough to read a text.

* Non Visual Information, which is the information the reader already knows. Non visual information can be of many kinds but some very important kinds of non visual information for reading are the knowledge of the language(ortographic -such as factible letter combinations-, syntactic and semantic as well), the knowledge of the topic, the knowledge of how to read, the knowledge of the graphic organization of a given kind of text, etc. But even the knowledge of a certaing literary style will help.

He proposes that for reading non visual information is more important than visual information and that the more non visual information we have the least visual information we need to read a text.

He explains that there is a severe bottle neck on the amount of visual information that the brain can recieve and process so what we really do when reading a text is to discard at least 75% of visual information and use our perceptive skills to predict and confirm the meaning of the text trough the reduction of uncertainty by the redundancy of the information contained in our brains (non visual).

So the main hability behind reading is to reduce the dependency on visual information by the elimination of choices trough redundancy of the non visual information.

Perception, memory and sense are the key concepts in Frank’s view of reading because they determine the low capacity of the brain to process visual information and the high capacity of the brain to deal with meaningful information.

Reading must be oriented to the discovery of meaning and not the decyphering of sounds or words because only meaningful information stays in memory to feed perception and to relate with our cognoscitive structure (already owned knowledge).

So this brings us to the medular topic of this thread:

¿Are we reading letter by letter, word by word or phrase by phrase?

The answer would be that we are reading the meaning of the text and to achieve so we would read at the nevel we need on any given moment to reduce our uncertainty of such meaning: words, letters, phrases, punctuation... even a single letter stroke, or a couple of letters, a white space between letters or the blurry shape of a word could help us to define our understanding of the meaning of a text.

The method proposed by Frank Smith to do such a thing is “feature identification” (hope I translated it right -análisis de rasgos-) which would only look for a given feature or a set of given features on an item to identify it (as a way to reduce the uncertainty along with the non visual information and the prediction of what could this item be).

Reading in other ways (on a pure letter, word or phonic level) would slow down our reading hability by saturating our short term memory with meaningless (unrelated) information. This is why meaningless or random text can’t be read fast, there is no way to predict or understand it.

So this is in a gross and unsophisticated review of the of Frank Smith’s theories on the reading process. I’m willing to clarify the topic with details of the author’s text and arguments if requested. Hope this doesn’t come out of place and that can be useful to the discussion.

I have done a second review of the thread, I have to admit I have not read it all (thick stuff) but I realized that the view expressed in my last post is similar in a broad way to the ideas Nick posted in his last post in this thread.

Héctor

Héctor, I think your post also chimes with a few other ideas discussed here and in previous discussions and debates: for instance, Peter’s emphasis on role architecture and featural recognition, which seems similar to what you have translated as ’feature identification’. The importance of non-visual information relates directly to what I have called readerability, i.e. the core competency of mature and experienced readers, which I believe is overwhelmingly responsible for speed and accuracy in reading, against the relatively minor impact of type design specifics.

While I think a lot of this is quite in line with my own stance. :-) Things like “discard at least 75% of visual information” is exactly the type of thing I put forth. That said, it’s important to try to separate the issue of readability into chunks which are manageable and of course pertinent to the people at hand... like type designers. For example, “non-visual information” is great, but it’s peripheral to type design.

And to yet again correct John’s stance at the beginning of this new year :-) I would remind everybody that no matter how minor an improvement is, it can be relevant to some people some of the time. Type design as a whole is (almost) never of critical importance anyway, so saying “it doesn’t matter” is self-defeating. The amount of effort that John for example -rightly- puts into the details of his designs is not of greater relevance than things like lateral masking, the positive role of serifs, etc.

hhp

Where type design matters most, in terms of readability, is almost always linked to specific reproduction or rendering conditions, and the better those conditions are the less type design matters to readability and, conversely, the more it matters in terms of aesthetic, cultural and stylistic concerns, in which mere variety is not the least of benefits. We have such a rich typographic heritage precisely because over 550 years we have developed methods of reproduction that render most distinctions in terms of readability between different text faces moot. If something is going to be well printed at typical text sizes, no designer wrings his hands and says ’Oh dear, I wonder whether Caslon is more minutely readable than Garamond or the other way around’. Typographic design decisions on the plateau provided by readerability, non-visual information and good reproduction conditions are almost never concerned with readability, and almost entirely concerned with aesthetic, stylistic, etc. considerations. It is only as the reproducion conditions are reduced in quality — e.g. sub-normal text sizes, poor paper stock, low resolution, etc. — that readability as an aspect of type design increases in importance. And in order to design readable text types for such conditions you need to know precisely what the conditions are. This is why a typeface that has been carefully designed to be readable on screen using a particular rendering system is not going to be a model of readability if used to print a telephone directory. Readability is totally dependent on output conditions. So where I disagree with Hrant is that he seems to want to make readability the primary focus of the type designer’s general work, whereas I see it as a specialised aspect of that work that is of relative importance depending on output conditions.

By the way, the specialisation of type design for reduced reproduction quality can lead to some interesting imbalances. Poor conditions tend to get lumped together, e.g. telephone books are printed very quickly on poor quality paper and at small sizes, so type designers making fonts for use at very small sizes tend to assume other conditions will also be sub-optimal. Mark Barratt gave a very interesting presentation at ATypI in Helsinki about selecting fonts for use in mass-mailed response forms, which are set at very small sizes indeed but incredibly well printed on coated stock. He found very few types that were appropriate to this task in part, I think, because most types designed for use at very small sizes presume poor printing conditions. If the printing is so good that ink traps and other readability enhancing features designed for less good reproduction conditions are distorting the letterforms even at 5pt, the font is not appropriate. Through this project, Mark ended up identifying a poorly supplied niche in the type market.

> ‘Oh dear, I wonder whether Caslon is more minutely readable than Garamond or the other way around’

But you -rightly- ask yourself questions that are even less “important”
in your own design work*. Importance is relative and subject to context.

* Including concerning readability - like in your SBL Hebrew face.

> he seems to want to make readability the primary
> focus of the type designer’s general work

Not at all. I just want a better balance than we’ve had since the 90s.
Actually though I shouldn’t blame it all on the 90s: chirography has
been impeding readability since day one.

hhp

Readability depends a lot on many factors other than the typeface, it’s even more difficult to get someone to do a good typesetting than to usea good typeface.

From what I have learned now on the topic I can tell that legibility is a thing that depends more on the reader (interest, habilities and previous knowledge), the syntactic structure of the text and the page layout than on the typeface itself.

Of course that the typeface would have a role to do in helping legibility by setting a good text rythm and color, and giving good character form definitions but being these factors covered I would think that a typeface would help legibility more by augmenting the reader empathy and interest for the text by the stilistic contextualization of it.

There is still a room for improvement in the performance of a typeface to aid legibility but also I think that it is not that much in normal printing conditions...

Héctor

Sure readability depends more on a bunch of other things more.

But: it still depends on the type too; and we make type.
This is really all the reason we need to care at least a
little bit about the readability of type, to care enough
to think about improving it. Worrying instead about
improving fashion and/or self-expression to me is
frivolous and selfish.

> I would think that a typeface would help legibility
> more by augmenting the reader empathy

Only if laymen cared a lot about “style” in fonts... at 10 point.
They don’t.

hhp

Including concerning readability - like in your SBL Hebrew face.

Well, the Hebrew script comes with its own built in conditional challenges :) — also, I was making a font that would be used a lot in teaching the Hebrew language, so making clearly differentiated forms was pedagogically important. Learning to read a language that has no tense structure is difficult enough, without students being confused about what letters they are seeing.

Only if laymen cared a lot about “style” in fonts… at 10 point.
They don’t.

Do you think all those millions of Word users who decide they like Monotype Garamond more than Times New Roman, or vice versa, don’t care about style in text faces? Style is exactly what they do care about, often to the detriment of readability. They like the look a certain typeface gives to the page, even if they can’t articulate why and don’t know anything about the history of the particular type or its place in the typographic scheme of things: they respond intuitively and personally to style. And style is exactly what motivates the legions of font ’collectors’ and ’swappers’: if they didn’t care about style, they wouldn’t want all those different fonts. And it is precisely because laymen do care about style, however ignorantly, that software companies have realised that bundling diverse font collections adds both perceived and effective value to their products.

”...they respond intuitively and personally to style.” “...even if they can’t articulate why and don’t know anything about the history of the particular type...”

Some may respond to style, others just see all those fonts in their list and try them just to be different than the last thing they typed. Some people may just choose a font based on where it is on the menu (proximity to default faces for instance). Computer users are used to their menu driven world. They make their choices based on what is on the menu, not what “style” they may truly like best if they shopped around. Readability is adjusted by them in the “Point Size” menu and their judgment is, more than likely, based on the screen image, not the printed page. They never use 10pt in their own word processing except maybe for footnotes. If 12pt were not the default size, they probably would use something bigger. When they see 10pt in a printed book (which as you know is quite common) they rarely respond to the style. Instead, they just read the text. Perhaps this is what Hrant means?
The major players in the bundling arena put in mostly readable fonts but add a few decorative and display faces too. I have seen plenty of lay users pick one of these non-text fonts for body text “just to be different.” There is nothing Microsoft, Adobe, or Apple can do to ensure that lay people use proper type for the job outside of not including any that don’t work for text in their bundle. That would be a marketing mistake given your “perceived and effective value.” (I get really tired of seeing all certificates done in “Chancery” and “Olde English”.)
There is nothing type designers can do to ensure good typography either. We can only do our best with the design and hope we don’t have to see uses that make us cringe too often.
ChrisL

Don’t give the dumb half too much work:
http://news.bbc.co.uk/2/hi/health/4723216.stm

hhp