W3C homepageWeb Accessibility initiative homepage

WAI: Strategies, guidelines, resources to make the Web accessible to people with disabilities

19 November 2012]

WAI R&D Symposia » Text Customization Home » Transcript

Transcript for Text Customization for Readability
Online Symposium 19 November 2012]


>> SHAWN HENRY: Welcome to the call. We're just getting logistics set up.

Everyone is muted for now. We'll unmute when it's time to talk.

Feel free to put any questions or comments in the chat.

>> DAVID SLOAN: Hi, Shawn. Yes. I think everyone is there. I just made a note. My name is David Sloan. I'm one of the co-chairs. You'll be hearing a lot of me in the next two hours.

>> SHAWN HENRY: Good morning, [afternoon, evening] everybody. Welcome to the symposium on Text Customization for Readability. This is Shawn Henry. And the symposium is put on by the W3C Web Accessibility Initiative, that's W-A-I. And our Research and Development Working Group.

So on behalf of the Co-chairs, David Sloan and myself, we would like to thank you to the Research and Development Working Group participates, to the Scientific Committee who reviewed the submissions, to Shadi Abou-Zahra who is the W3C staff contact for the Working Group. Thank you to the authors who submitted papers. For those who contributed through e-mails and will continue to contribute. And then thank you for everyone participating on the symposium call today.

Our R&D symposium have four main parts. The first one is we define a topic. Then we solicit paper submissions, which are peer-reviewed and then published. The next step is the symposia discussion which is what we are doing today. And then finally, we write a research report which the entire Working Group contributes to and is available for, a draft is available for public review.

That is where we are going forward here.

As a reminder all the participants on the call are expected to have read the papers beforehand. The discussion today is to further explore issues and expand our understanding of the topic in order to contribute to the research report.

A few logistics. The call itself is not being recorded. There is live captioning and the transcript will be put online. The chat will also be available online. You are encouraged to put brief comments on the content of the discussion in the chat. That will be available online.

We have a publicly available e-mail for specific comments and we'll mention that at the end. If you have additional information after the symposium that you would like to share, we will encourage that at the end of the call.

For the main part of the symposium today, Dave Sloan will lead us through the agenda. He will ask paper authors as wellas some of the e-mail contributors some specific questions. We hope to have time in each section to open it up for any participant who has additional questions for the paper authors or comments on the topic to speak up. If you would like to get on the speaker queue for a certain topic on your phone keypad you press 4-1-#. 4-1-pound sign. That will add you to the queue, the list of people who want to speak.

You can also type your question or comment in the chat as well.

That is the logistics. That's the plan. As a reminder, today's topic is very specific. The topic of Text Customization for Readability. We are exploring the needs of people with low vision, dyslexia and other conditions, and even in other situations that impact reading.

The focus for today is specifically on text customization requirements and functionality. That is providing users the ability to change or customize the text display for their particular needs. That's what we are going to focus the discussion on today.

With that, I think we are ready to get into the specific agenda. I will hand it over to Dave Sloan. Dave?

>> DAVID SLOAN: Thank you very much, Shawn. So we split today's symposium into three main sections. Firstly, we would like to spend about 30 minutes or so discussing text customization needs and requirements in terms of what we already know, what research has been done to help us understand what specific requirements are appropriate for specific groups. Moving into that in a little bit more depth, discover what are necessary requirements and what are perhaps advisory suggestions that will also open up the debate between requirements that may be diagnosed versus preferences that might be just somebody's subject I have opinion on what -- subjective opinion on what seems to work best. We'll talk about that as well.

Then we'll move into the idea of dividing text customization content into the Web environment, whether it is through user agents, assistive technology, or through Web content itself which is at the moment one route to supporting customization. It may not be an appropriate route but it's something to discuss.

Finally we will talk about what we need to do going forward, what further research is required.

[Understanding text customization needs and requirements]

>> DAVID SLOAN: Going back to the first stage in terms of understanding text customization. A number of our authors give submissions that explore this area in some depth in terms of looking at specific needs for specific groups.

I am going to ask our authors are in turn to give very briefly an overview of where their submission explored specific requirements for specific groups. I would like to start, I had Shawn at the top of my list but I am going to move down and ask Luz if you could explain very briefly what you found from your research that helps people with dyslexia and perhaps elaborate in telling us how your finding us might support or refute the concept that there is not a single color combination that will meet most users' needs.

That question is to Luz first.

>> DAVID SLOAN: Luz, can you tell us about your research in terms of what it found for people with dyslexia?

>> LUZ RELLO: Yes. So our results do not support that there is only one optimal goal or combination for reading. We tried out eight different color combinations and our results support with the data, previous recommendations. For instance, the recommendations from the -- association of dyslexia or from Peter Gregor and Allen Eulle's work. That supports that, for example, not very -- not high contrast between font and background is recommended. In terms of colors, we have preferences and they vary among users.

>> DAVID SLOAN: Okay. Thank you very much. I think moving on to the implications in a short while. I am now going to move on to ask Wayne to briefly elaborate on the observations you made in your submission explores customizing text at the element level. Can you tell us a little bit more about that observation?

>> WAYNE DICK: One thing that I noticed was a need for equal access in terms of getting all the information that a normal reader gets. That includes things like whether something is a citation or whether something is code. There are subtle type graphic indicators that are usually represented by HTML elements that are given different styles which ordinary readers see. But which most assistive technology at this time, either enlarges in a format that is not friendly to the user or simply collapses it all into one style, losing the semantic indicators. So I studied this. I believe that probably the finest granularity we can get is the element level, the document element level. And that that is probably enough if we look at HTML. I'm not certain if you can actually get down to stale, different styles of different text in PDF. I don't know if their tag structure is refined enough for that.

>> DAVID SLOAN: Okay. So if I understand what you are saying correctly, you are saying that this same HTML element might be presented in a different way depending on where it appears in the Web page and, therefore, somebody who needs to customize text wouldn't appreciate the subtlety in difference. Therefore, that customization has to be supported?

>> WAYNE DICK: That's right. That customization would not give them the full information to the document. When a person looks at a document that has been formatted, say, for the American psychological association or any professional organization, they can build a mental model just by looking at the typography. The paragraphs look a certain way, headings look a certain way, lists look a certain way.

Even finer, a citation of an article may be italicized or an article name may be italicized where a book name may not be. Those kinds of indications are usually missing in text customizations that exist today.

And I claim that that really is necessary for a person with a print disability to have this same set of information that the normal reader has.

>> DAVID SLOAN: Thank you. We'll come back to that shortly. For now, I would like to ask Anthony to share with us your perspective of readability issues related to typography and subtext. Anthony?

>> ANTHONY LEE: So I think the problem for me with sub-pixel rendering is almost the typography or the font choice becomes relevant because I can't look at anything, any text whatsoever which has got, been rendered using this form of technology.

So a good example would be Internet Explorer 9. There is no ability for reasons which I really have nobody understanding, Microsoft have decided that you cannot turn off what they call clear type in that browser. So I cannot use that browser at all.

So that would be would be the issue for me. I don't even get into a discussion over font sizes or font types. I just can't look at any of it. And as I understand there are a number of other people in the same boat as me.

>> DAVID SLOAN: For clarification, could you give a rough figure about the percentage of text that you encounter that is affected in this way? Is it Web content? Is it the user interface of your browser? What is the extent of this problem?

>> ANTHONY LEE: For example, I'm -- at the moment I'm okay. For example, if I use Internet Explorer, I use Internet Explorer 8 with clear type supplied.

So broadly speaking now at the moment until Internet Explorer 8 goes away, I am okay. The difficulty I have is really with the future because Microsoft likes people to stay on its upgrade path. If it has decided that it will not be possible to turn off sub-pixel rendering going forward, then it will become an increasing percentage of instances where browsing using Internet Explorer, which is necessary in the corporate environment, will just rule that out for me. So it will be 100 percent in time of stuff that, my inability to use the Internet where Internet Explorer is required. It does play out in other ways as well. So, for example, PDF for example or images on websites these days are typically rendered using sub-pixel rendering. And once that has happened there is nothing that the user can do apart from zooming it to change that visual effect.

So there's a lot of stuff these days on websites that I have to sort of not look at. I just focus on the HTML text and look away, which is pretty annoying. And the other one is PDFs are used all the time now and these days most text that is embedded into a PDF has been rendered using sub-pixel rendering in the first instance. Once that's done it makes it very difficult for me to read it without getting a sort of sickening headache.

Some PDF readers are better for me than others. For example, I struggle with Adobe. There is another one I can't remember, one off the top of my head that I use. That will come back to me later in the call. But at the moment the situation is sort of under control, but getting worse. That would be the best way to summarize it.

>> DAVID SLOAN: Okay. Just to quickly feedback on one thing. Olaf has indicated in IRC chat that fonts can be eliminated in Adobe Acrobat. I don't know if that's all versions or if it is introduced recently. That's something to share at this point.

>> ANTHONY LEE: That's good to know, but once it has been subset into a document with sub-pixel rendering, is it not set in concrete? That's something I can look at later. I think the interesting thing for me, one of the papers talks about if you have a tag PDF, you can read it through HTML and then you have a lot more control over it. The solution to that problem may lie somewhere along those lines. I just don't know at the moment.

I'll look into that.

>> DAVID SLOAN: Okay. I'm going to quickly read out a question from Clayton Lewis. Clarification question: Are these issues connected with text magnification of viewing of text default size as well? Is there a difference if you change the size of the text and the impact?

>> ANTHONY LEE: Is the question for me?

>> DAVID SLOAN: For you, Anthony, yes.

>> ANTHONY LEE: So basically the sort of what I call the clear type effect, because I'm mostly a Microsoft user this applies the text as all sizes. The worst visual disturbance I get is when the text is really small. I know that's counter intuitive because it is designed to make it better when it's smaller, but to me it intensifies the visual disturbance.

If you make the text bigger it becomes readable but it looks a bit funny around the edge, which again I find it quite hard to understand why the industry has gone down the route of sub-pixel rendering.

For example, if you look at text on a mobile device where sub-pixel rendering doesn't touch because it has a rotating screen. If you zoom, for example, HTML text on a tablet, it looks absolutely fantastic to me. But the short answer is the smaller the text, the worse the visual disturbance, the worse the headache, the worse the sickening feeling.

>> DAVID SLOAN: Thank you, Anthony. Can I ask Eileen to reflect on how text customization helps the people you work with? So this question is for Eileen.

>> EILEEN RIVERA: Hi, there. My comments would be from a lay perspective, not the research side at all. But from my interaction with colleagues who have low vision, we encounter similar issues.

A lot has to do with the fact that just the technologies that are presented are sort of crude in that they try to just enlarge everything at once. When that happens you don't get the full landscape of what information is before you if you can imagine reading half of a sentence and scrolling to the other side to get to the other half, you won't get far in the paragraph to understand what you're reading. It's painful and laborious. If I could have what is on my wish list, everything text would free flow on the field on the screen so there would be no need to go off the screen and scroll to the left or right to read what I need to read.

I think that was one thing I could have for Christmas, that would be it.

The other thing that I find problematic is I use a lot of the inverse contrast settings just for my Microsoft applications in Windows. But I believe that those settings, which I think many people use, who don't use the full assistive technology packages, they use inverse contrasting, things like that, or high contrast color. I don't think that is ever tested with the.

>> EILEEN RIVERA LEY: When those schemes are applied the website is unreadable. I would have them at least test the websites with those kinds of settings.

And I think those were the biggest things I can imagine. I think there is a huge awareness issue. Lots of ordinary users are not high-tech and I put myself in that category. They wouldn't know the first thing about how to use the style sheets or if I can the problem themselves. We have to be much more user-focused and, when one size fits one, we all have very specific preferences and needs to optimize our efficiency.

So I guess I would like things to be that customizable, especially depending on what the settings are, what tool you are using, where you are reading, reading on the bus versus in an office or some other place, I think we use different customized settings for each application.

And on the website, depending on how you are looking at it or how you feel that day visually.

>> DAVID SLOAN: So there's an issue of people's need to be even within a single website.

Thank you, Eileen. Can I ask Shawn to reflect on your research looking at user generated style sheets and the adjustments that people make when they write user style sheets to customize display of Web content?

>> SHAWN HENRY: Yes. I have actually done research with a couple of different approaches to looking at what people customize in text display. And the one reported for this symposium was looking at CSS. The things you were expect were indeed some of the top things that people customized, font size, font family, color including background color and text color.

The one that was high that was a little surprising was line height, how many people did customize line height to increase the space between lines.

Then there were other things, text decoration, text alignment, font weight, margin and then some other things that were also customized although less frequently.

Then the other aspect that I wanted to mention was that in several of the style sheets people had customized text differently at the element level. For example, people had changed the headings to be different from the main body text and had changed the text of links. And even changed different IDs or classes different bits of content on the Web page to be different as opposed to just making a global change to the whole Web page.

That's a brief summary.

>> DAVID SLOAN: Okay, thank you. Again, we are hearing the granularity of change that people make seems to be quite fine-grained which supports what we have been hearing earlier.

Could I now ask Whitney to share with us the similarity in customization needs that you found in your work with different user groups?

>> WHITNEY: We were actually working on a large university website and our concern was not with the accessibility features of it per se, but with trying to decide what to make the default text and the default presentation of the pages so that it would be usable and accessible for the broadest number of people before anybody had to apply any customization.

This university is a large mostly distance-learning university. So it has some rather special audiences. It has the usual teenagers who are the usual college audience. It also has a larger population of older adults, people with disabilities including learning disabilities, reading disabilities, excuse me. And also people with low English proficiency. The challenge that was posed to us was: Could this be done with a single website or were these audiences so different that we needed to write differently and to present the information differently?

When we looked across research including some work by Ginny reddish who I understand is on this call, with older adults, what we found was despite the audiences were described as having very different needs, the guidelines themselves were quite similar. For example, all of the guidelines stressed the need for larger text size. The definition of large was different, it was over 14 for teens and over 14 points for low literacy.

Another thing we did was look at images and suggested that images need to be meaningful for all of the audiences because they were easily distracted or in various different ways. But the most important one, I think, was about where several guidelines in all of the sets of guidelines about how the information was presented. Not just how it's written, although plain language appeared in all of them, but that it did something that we summarized as calling it break up walls of words. The presentation of the text visually needed to have clear headings, needed to have short paragraphs, needed to have good spacing to put things that were lists into lists so they weren't just big blocks of texts.

That the headings not only needed to exist but also needed to be meaningful text. And that the text structure needed to progress from what we called the shortest simplest answer to more details. So that it would be easy for someone who was impatient like a teen to read the page and get the key points. That it would be easy for someone who read everything, like older adults, to understand the structure of the information.

And for people with various forms of reading problems, that it was easy to understand whether it was worth the effort to read deeper into a section.

>> DAVID SLOAN: Okay. Thank you very much, Whitney. I should clarify that Whitney's contribution was received by e-mail. So it will be available through the public comments archive linked from the text customization symposium Web page on the WAI site.

So we have had a good discussion or description of the different needs of different groups and how some of those needs are very similar despite, or the solutions to accommodating those needs are similar despite the differences within the groups. I will come back to the contributors to discuss the work that might be required either to educate designers and developers or to adjust existing best practice shortly.

But for now one of the points that emerged from the discussions really related to the relationship between required needs for people to be able to read effectively and for optional suggestions. And I wonder if I can ask -- let me ask Wayne in terms of element level customization, would you consider this an essential need or a desirable need? My question is to Wayne.

>> SHAWN HENRY: Dave? Do you actually want to take the queue for the previous question or do we need to move on?

>> DAVID SLOAN: There are a number of questions that have appeared in the IRC log. And I think I would like to come back to them before we move on to section 3.

Let me take the question from Gordon, please, first before we move on and I'll come back to the other text questions after that.

So is Gordon available?

>> GORDON: I'm a person with low vision. I'm also a researcher on vision and reading. Speaking from the perspective of a person with low vision I just want to echo Eileen's comment. I use configuration, a Windows configuration with bright text on a dark background. There are many websites that are just incompatible with that. Things disappear.

There's a real problem there for people who use this sort of contrast mode. I did have a question for Wayne about the element level an sis which I think is very interesting. I think the question is this: There might be a lack of congruence between I think what Wayne is referring to as the author's semantics and the elements. So an author might not be consistent in the use of fonts or whatever in association with particular types of elements and vice versa. So I'm wondering where the real basic level analysis should be. In the HTML elements? Or the author's semantics? Which should be the real focus?

>> DAVID SLOAN: I'll put that question directly to Wayne. Wayne, would you like to answer that question?

>> WAYNE DICK: Very much.

I thought about this. My proposal is nowhere near perfect at this point. I'll throw this out, it needs refinement as to things that need to be considered.

I settled on the element because it can be determined by program. And the semantics associated with HTML elements are defined by the HTML language.

If someone comes along and they say I have a heading level 1, we would assume that the author used that heading level 1 appropriately. If the author used it appropriately, then it would make sense to make all heading level 1s look the same way so the user can know that they are in fact running into a main heading or second level heading.

For example, the way I do it for myself, I do have low vision. I make headings the same size text as nor mall print except I make it in a different font face and I put a sequence of dots in front to tell me what the heading is. Level 1 heading is one DOT, the level 3 has a space and a dot for a level 4.

I think that is sort of my compromise between what can be computed without artificial intelligence and what is actually out there in data.

Does that answer your question?

>> DAVID SLOAN: Thank you very much, Gordon. Do you have any quick feedback on that answer?

(There was no response.)

>> DAVID SLOAN: I'll move on. I notice from the IRC chat that there is a lot of discussion about readability. I would like to come back to that when we talk about existing text functionality and interfaces to talk about that as a solution and a possible source of further data.

So I will come back to that shortly.

We were beginning to talk about the relationship of necessary customization requirements and optional suggestions to improve readability. I would like to ask Luz to elaborate on your research, and in particular the relationship between reading performance and personal choice. If you could begin by reflecting on the difference that you found or the relationship you found between reading performance and personal choice and what the implications might be if optimal performance is not achieved by a preferred option and what that means for preferred option.

>> LUZ RELLO: Thank you, David. Yes, we found correlations, mathematical correlations between personal agreement and reading performance. So since there is no agreement between both, we believe that both need to be taken into consideration when developing standards and guidelines.

So far, at least for dyslexia, mostly only personal choice of the users was taken into consideration for recommendations. So I believe we shall include also recommendations based on actual reading performance when building standards and guidelines.

>> DAVID SLOAN: So would you suggest then that a user agent should somehow suggest to a person an optimal customization set, for example, text color and background color because this has evidence behind it to suggest this would help readability, into that selection?

>> LUZ RELLO: Exactly, this is exactly what I meant. Of course, also ideally the user can customize according to their preferences, but yes, this is what I would suggest.

>> DAVID SLOAN: Okay. Thank you, Luz.

I would like now to come back to Wayne and to ask a question relating to the other submission that you put to the symposium relating to the typometric X tool which began to explore the idea of diagnosing a Web user's particular requirements and presenting customization based on a more formal diagnosis.

Wayne, can you talk about that in a little bit of detail and to explore this issue between customization, a diagnosed customization and a preference?

>> WAYNE DICK: Yes, a lot of things guided me to that and the paper that was discussed just before helped me think about that.

My basic idea was that we don't, there's two things that we need to know about. There's reading performance which is kind of a measure of perception in that you measure reading speed and fixation time. But one of the problems that people with low vision -- and I have low vision, too -- experience is extreme fatigue. Partially because we read slowly, we have to read longer; but also because of other factors of the typography. I would like to simply ask the question: What do people choose for their typography? And then would that choice, and give them every single possible stylistic option they can have. And with cascading style sheets we can make all the changes that Shawn mentioned, but we can also play with the spacing between letters. And not in gross ways but down to one pixel larger and that would help people get exactly the typography, what I think Eileen was talking about, one size fits one for each person to choose the size that fits them.

Also you can pick a size for when you wake up in the morning with 20-100 vision and go to sleep in the evening with 20-200 or 20-400 vision and you can have another style setting for that as well, but it will be users.

Once we gather enough people with their style settings, we could check the standard parameters such as reading speed, how that compared with what is called their critical print size which is the smallest print size they can read at optimal speed. And those kind of factors to understand what the difference is between preference and maybe what would make the difference between ar person being able to read enough to keep a job and not read enough to keep a job.

So I basically am building this tool so that we can actually do testing on that at this time.

>> DAVID SLOAN: Can I ask what your view might be on how frequently this tool would have to diagnose a user's needs? You are suggesting there's a diurnal change as well as a change from one website to another. How frequently, what is the balance between the detail of the diagnostic process and the frequency that would have to take place in order for it to be useful.

>> WAYNE DICK: I think I will be aiming at nice text at this time, like coming out of book share.org or things like that that doesn't propose a lot of problems like having to reposition elements and stuff like that to begin with, so we can sit down and focus on how people read realistic text when it is in a format that they can use.

Once we kind of get that kind of determination settled, you know, I think that then we can really talk about restructuring programs to put Web pages in a form so that everything word wraps nicely and all of that stuff and on top of that, the person has their one size fits one process.

Now, how often would a person have to do this? My guess is about as often as a person would have to change glasses.

I think it would be very much like doing an optometrist change, like I have four style sheets I use on a regular basis so that I can read. I use one for composition that has slightly smaller print. I already know what I wrote.

I have one for reading which is larger and in a different color scheme because I find that more comfortable. I have one that I read at night because I'm really tired and it's kind of, it gives me exactly seven to one contrast with the minimum light I can use because by the end of the day my eyes are sore.

And I have one that I use on general Web pages that I call gentle which doesn't do any reorganization but changes color so it's easier for me to see. I have a feeling that people might have like a spread like that, but when I change those, I maybe change them once a year, once every two years. And that's the frequency answer. Sorry to go so long.

>> DAVID SLOAN: That's okay. Thank you very much. Lots of interesting observations there and again some interesting observations on IRC channel, particularly this debate over the extent to which individuals who can benefit know what is best for them and the role of perhaps some form of automated performance evaluation to identify the best customization combination I would like to ask Suzette Keith who submitted something by e-mail to share with us her perspective on line spacing and whether adjustment of line spacing would be considered an essential or desirable customization requirement. Suzette?

>> SUZETTE: Yes, I have -- I have something that helps me -- something that helps with the effects you get from having cataract surgery. I use increased font size because it is a better size for the position of my relationship between my reading distance and the screen. But then I find the blocks, large blocks of text just become a great blur that is very difficult to skim read or to read thoroughly.

So having mentioned my problem while I was at the meeting, they kindly developed a booklet which I compressed, which puts a little bit of air between the lines and this is, it helps because it has the increased line spacing. I need a package. I need to adjust the word lengths at the same time and adjust the gap between the lines and that package together makes my viewing a lot more comfortable.

>> DAVID SLOAN: Thank you, Suzette. Do you consider that as it exists right now a browser extension? Would you consider that a piece of fundamental functionality that all browsers should support?

>> SUZETTE: I think it needs to be really easy to use and get to. I mean, I'm also doing, teaching older people to use computers at the moment. And even getting them the mouse as far as the zoom commands is really too complex to take on when you are a new user.

Obviously you can do the keyboard short cut, but that's asking a lot for seniors. I need it just for very particular situations. And I just want it as a very quick switch on/switch off. I don't want to set a whole style sheet. I'm not into changing absolutely everything. It's a minor adjustment and the booklet or little toolbar would suit me fine.

>> DAVID SLOAN: Okay. Thank you very much for that.

I think for reasons of time I would like to move on. We may welcome back to discussing gaps in knowledge of users' needs for customization. This was a topic that didn't emerge in the submissions to any great extent. So that is something we may well decide to discuss at the end of this symposium if we have time. Observations via IRC or request to speak in that area would be welcome.

[Integrating text customization functionality and requirements]

>> DAVID SLOAN: Now I'm going to move to the second key area of discussion which is integrating text customization functionality and requirements. First of all, talk a little bit about the extent to which existing text customization functionality and interfaces supports users' needs.

Now, we have heard a fair degree of coverage of this area based on what is already useful for people and what is not useful. I wonder, Shawn, if you could share with us from the user style sheet perspective as one solution to supporting customization, what evidence do you have of usability issues in the authoring and applying of user-defined CSS in different situations?

>> SHAWN HENRY: Yeah, the first thing I would like to do is to clarify that the research I did was, the goal was to figure out what aspects of text that people do need to customize. So this was one study. I did another study that looked at how people customize text in a word processor and what they try to do with PDF documents.

My focus was not so much on how people use style sheets per se but what we can learn about how people use style sheets.

Having said that, just to answer your question, certainly there's ample evidence about the difficulty of using user style sheets in current user agents that there is some, I think in the paper there's a quote from one of the users that was something about how much work it was to undo the mess that people made with their style sheets. So how hard it was just to get it readable.

But in addition to that, just the fact that there are so few people who do it and that in both studies I found people who customized text in are their word processor but they don't know how to do that in a browser. That was a significant issue. Even recently on a Web accessibility mailing list there was someone asking how to do that. So I will look forward to talking later in the symposium, but my position is basically that they shouldn't -- that users shouldn't have to have the knowledge and expertise in designing their own style sheet in order to customize text.

>> DAVID SLOAN: Thank you very much. I'm going to take the chair's liberty of adding a personal observation which I submitted by e-mail earlier today to agree with you and the diversity and sometimes unexpected ways that people will apply changes to text appearance. I already see in the IRC chat some participants in the symposium reflecting on situations where they see people leaning towards the screen in order to accommodate readability difficulty. In our research here in Dundee in Scotland we have seen instances of or heard of instances of people copying and pasting hard to read text from a website to put it into Microsoft Word and then apply style changes there in order to read it. Because they know that in Word you can control how the text appears and our assumption is that for some people the awareness of customization support within a browser is very low. So their answer is to move the text from one application to another in order to make changes we have an interesting submission from Silas Brown who is not on the call today but has done some work exploring user style sheets. We unfortunately don't have time to go into his submission in any detail, but it is referenced in the public comments e-mail thread. Accessible via the symposium Web page. If you want to read that submission, I direct you there.

I will summarize it and say that he discussed the limitations in the specificity of user style sheets which I think is similar to Wayne's point of element level customization.

The impact of applying a user style sheet on scripts that control appearance and that are very dependent on using the Web authors' style sheet. Silas Brown also talked about user control over display of other Web content that may not appear by default. So, for example, the title attributes of a particular element that may only appear when the mouse pointer hovers over the element in question.

So I would now like to ask Olaf Drummer on customizing support in PDF readers and his approach to addressing that situation.

>> OLAF DRUMMER: I would briefly like to comment on the status quo before we did our project. That is, there were very few useful mechanisms to customize rendering of PDF content. You can do a reflow in Acrobat, but that's conceptually broken already from my point of view at least.

You could apply contrast settings or color settings, but in both cases this would only work for very simple PDF files with a immediate break for anything that is more graphically interesting.

I have run into users who use extraction tools like PDF to HTML, which is a tool a couple of years old that they grab the content from PDF files and reformat it as plain text or HTML. But forgetting all the semantic structure in the file. I found this situation very useless, so to speak. Because actually working on PDF and accessible PDF for the last two years or so, working on the PDF site, I was looking for a tool that could give me a more intuitive rendering or presentation of the PDF text content. Because there was none, I decided to make one using some of the engineering resources in our company.

We started doing this project in August and worked on it for about two months to end up with a prototype.

I think we came up with what is the first ready to use tool to turn PDF into HTML reflecting the semantics that is in the PDF file. And you will only find such semantic if the PDF is a tagged PDF file. Our tool will not work well on tagged, and it will not do a good job if the PDF is not tagged well.

And just to maybe end my summary, the result of our export from PDF to HTML is simply opened on the default browse other the user's system and from there you do whatever you do inside the browser. You could apply user style sheets or just play around with the tools, like in Safari you can say increase text size as opposed to increase everything on the screen.

So that was not the focus of our work.

>> DAVID SLOAN: Thank you, Olaf. Just for clarification for the symposium attendees, your planned evaluation for this work, what do you plan in order to gather data on the efficacy of your approach?

>> OLAF DRUMMER: We are enabling something. In terms of enabling something I believe we have been successful because we can transform the tagging information in the file into well-formatted HTML. Depending on the quality of the tagging you have well formed HTML with the headings and the tables and whatever you have, in the tag PDF file.

So we didn't do any evaluation of text customization as such. We kind of just evaluated whether you could go from a tagged PDF file to HTML in a useful and meaningful way. I think we can probably do that. By the way, the tool is available free of charge from our website as mentioned in the paper, beginning today. So we just released it today. Everybody can play around with it. We would be interested in learning more about how people use it and would also take in feedback to develop the tool further. At the moment we don't do anything in helping users helping with text customization as such.

>> DAVID SLOAN: Thank you very much. I'm not sure if Vasile, one of our authors is available to talk right now. I know he had problems earlier.

[Vasile was not connected to audio. He typed his comment and David read it.]
>>Vasile Topac: Although this is not my main focus I try to cover the customization of PDFs in www.textforall.net project. I try to convert PDFs to HTML with my experience with PDF to HTML conversion. I used several SDKs and tools for this. The main issue is the layout of the text in the HTML is very different from one in PDF. This can have an impact on the logical structure in the PDF structure.

>> DAVID SLOAN: So I think we can see that there is a challenge in exporting PDFs and content to HTML in order to support more effective customization.

In our discussion of text customization functionality and requirements, a number of participants mentioned readability as an application that has been quite successful and I wonder if I can ask, a number of people recommended it. Perhaps I could ask Whitney to share her perspective either as a user of readability or somebody who has done research on its effectiveness. Whitney?

>> WHITNEY: Yes, hi. I'm afraid I haven't done any research on it but I am a devoted user. One of the most important things it does for me is that it clears away everything but the content text. So I get a clean screen without competing distracters around the edges and I can set margins and color and line weight by picking a style. So I find it sort of the screen version of printing out a clean copy.

>> DAVID SLOAN: Okay. Thank you very much.

>> WHITNEY: And I think we haven't actually talked about distractions around the edges of the text. We talked a lot about the text itself.

>> SHAWN HENRY: Right, because that's kind of --

>> WHITNEY: How things are set. That's a bit out of the scope.

>> SHAWN HENRY: Right.

>> WHITNEY: I have had readers in projects where we were testing. One was on a site that I thought was actually quite distracting in its frame. But people talked about how readable it was simply because they had kept the content area clean and well formatted with a nicely large-sized font and good line spacing and a simple presentation.

So I think that effect of people being able to zone in on the part of the screen they want to read supports the notion that if we can get the text readable and clean and in an appropriately presented way, we will have gone a long way.

>> DAVID SLOAN: Okay. Thank you very much. Shawn, did you want to add something?

>> SHAWN HENRY: Yes, I did actually. When we started discussion of this, I actually had some talks with the folks at readability. I didn't pull my notes out from this, so I'm going from memory. If I recall, they were focused on broader issues of getting rid of the material around the main article, for example. And not as focused on providing additional text customization. So I look forward to later in the symposium when we talk about how can we communicate users' needs tools such as readability or some of the ones that we can look at because from my initial discussion I think they were very open and interested, but currently that was, my understanding was that was not a high priority for them to provide additional text customization functionality. I look forward to talking more about how we can communicate that need to tool vendors, developers.

>> DAVID SLOAN: Okay. Thank you very much.

[Moving forward]

>> DAVID SLOAN: I think at this point we are probably ready to move on to talk about the where do we go from here? There are lots of emerging issues in terms of what text customization means, the extent to which it should be supported, some other issues that we may wish to consider in this discussion, is the relative responsibility of Web content authors in users themselves and the manufacturers of user agents in collectively making it as easy as possible to apply changes appropriately and effectively.

So one general question we would like to ask and I'm going to ask one or two of our authors first for their views on this -- is how we might increase awareness amongst people who develop standards, policies, browsers, other user agents of the need for text customization and its benefits.

I wonder if I can ask Eileen for your perspective on how we might go about increasing awareness of the need to support text customization?

>> EILEEN RIVERA LEY: Okay. Well, I had an idea that was implemented on another project that I worked on but it was a contest around people who had diabetes about what they wanted the industry to do to make their life easier. They submitted, each person submitted a three-minute YouTube video explaining their situation.

Then they created a collection of those. But I thought that was a really creative way to sort of personalize the needs and help the industry understand the opportunities for better serving the customer.

So anyway, I would throw that out as a possibility.

Just basically I think if the technical professionals could spend time observing the kind of gymnastics that people with low vision go through to use a computer each day, it would build a lot of compassion, understanding and motivation for improvement.

>> DAVID SLOAN: Okay. So you are suggesting that making people more aware of the impact of the inability to customize text effectively might increase the profile of the need to accommodate this?


>> DAVID SLOAN: I would now like to ask Olaf, who left a very interesting comment on the IRC chat, to talk about his view on the move towards the mobile Web and the emergence of techniques such as responsive Web design and what the impact on supporting readability might be. Olaf, could you share your thoughts there, please?

>> OLAF DRUMMER: Sure. So I think it might turn out that mobiles are our best friend when we talk about text customization for people with low vision and other people as well. Low vision people are small, but mobile is a big market. A lot of stuff going on there. Recently the term responsive Web design has become very important. That means that your website or your content does adjust to whatever font effect your device has and the rest Lewis of the screen it has. Because they are all different it has to be very flexible if you do it right. If it's that flexible, it is also I believe useful for any customization interests.

And we may want to look a lot more in that direction because the only thing that we have the authors for is if their content is suitable for responsive Web design. In terms of the project I did in PDFs, it is so obvious that if you have a letter size or full size document and you want to look at that document on your smart phone with, let's say, a 4-inch screen, it is not funny. I mean, it was the -- even with regular vision it is not funny, let alone low vision or anything else. If the document is important to you, if you can reflow it in a semantically meaningful way so you can look at it and understand it without too much pain. Instead of jumping around on the page on the letter size full document, you get a flow of its content with adjustable text size maybe and just read through it piece-by-piece. It would still be less efficient than on your desktop because it's much easier to jump around down and up and right and left on the page of the PDF. So my dream would be whenever you look at a PDF document on your smart phone there would be a button that says just do a semantically meaningful free flow. For PDF this wouldn't be difficult as I showed with our project. We don't have any plans to come up with a implementation on mobile devices, but that is actually what I'm hoping for. I think we could learn a lot from this from mobile and about reflowing static PDF content for other use cases as well.

>> DAVID SLOAN: Thank you very much, Olaf. I'm distracted because I lost my connection to the IRC server. Carry on and hopefully I will get my connection back shortly.

Perhaps we could move on to talk about the issue of redressing the balance, I think, between browser customization support and efforts that individuals might have to undergo in order to write and apply their own user style sheet.

Wayne, I wonder if you can share with us at your point your views on how element level styling might be best supported to minimize the effort of an individual on having to apply their own preferences.

>> WAYNE DICK: Okay. Yeah, well, as you and many of you might know, Silas Brown and I were some of the only people who used style sheets many years ago. That's because they were pretty darned hard to write.

Obviously, people need a user interface to help them just choose their style in their own terms. You know, to actually look at a page and choose a font face and then choose another font face and key trying until they can do it. They might start with something as simple as do you want light on dark? Or dark on light? And then adjust the background colors, find themselves a good standard running font size and then maybe element by element specify how they want that to look.

It would be very much like a preference profile, and there are national and international solutions that are being suggested where people could have a profile of their disability needs so that when they had websites that could automatically adjust, that's being done in Boston at this time.

But the idea being that a person could literally through a graphical user interface choose how they wanted their page to look and then have that automatically generated. I think that's the only way that that can work.

Now, the point is once they have a style sheet, that doesn't mean you can't have input into like Microsoft Word. It would be very easy to transform a style sheet into a Microsoft Word template that a person could use in all Microsoft Word documents. If it was easier to work with PDF, it might be just as easy to have that base template that could be sent to them.

I think that the real question here is not style sheet but the ability for a person to choose a style based on their own sight and perception and reading needs. And then have that translated into the format needed for the document that they were reading.

One of the problems that occurs on the Web is that because pages are positioned all over the place and they are generally built for people with normal vision, even the standard of 200 percent enlargement which the W3C requires, fails in probably 50 percent of pages. Usually one column over writes another column, or something gross like that.

There also then is a sort of structural thing where pages need to be restructured in a logical sequence so that they can fit on a screen once a person has restyled it. I kind of see these as the essential manipulations that are needed. I think that's about it.

>> DAVID SLOAN: Okay. Thank you. Can I also ask Shawn, who had a question, to ask -- well, Shawn, do you want to read out your question or will I read it out for you?

>> SHAWN HENRY: I'll read it. So one of the questions I had, Wayne, on your paper, the typometric RX, describes this as a tool to use for diagnosis and study. And I'm interested if either the tool itself or a different version of it might be useful for users to create their own style profile.

>> DAVID SLOAN: Wayne, would you like to answer that?

>> WAYNE DICK: Okay. One of the outputs will be a style sheet. One of the first outputs will be a style sheet. We'll probably also create, for those people who are not as technical, there's a language called JSON which immediately feeds into the Ajax level, the interactive level of pages so that they can pull that in and apply those styles directly.

So our idea is to start out with a couple of standard styles, especially focused at HTML. I have a couple of friends who are trying to get a pretty good reader together for reading book share books. That will be our initial focus.

But the idea is to truly come up with a style or what I would call a data format that would be generalizable and we could transfer to any input that renders text.

>> DAVID SLOAN: Thank you, Wayne. There's now a question from Gordon. Gordon, would you like to give your question?

>> GORDON: Yes, really it's a follow-up to Wayne's comments. I like the idea of a preference profile, but I have a couple of concerns. One is it would certainly be multidimensional. There are quite a few variables here. They may interact with each other. The obvious one being if you choose a larger font you are going to have less characters per line on the screen. So I'm wondering how you feel about that, Wayne. It is pretty challenging to get a profile that will work globally, let's say.

>> DAVID SLOAN: Wayne, would you like to briefly respond to that?

>> WAYNE DICK: -- developed Web Adapt To Me was Vickie Hanson. And we actually tried this out with a few people at our university. And there was quite a bit of variance. What we found was that for some people it wasn't very effective. And for other people customization was quite effective. The variance was not only in size where you have people who range from 25 words a screen to 200 words on a screen. But also just the contrast. We had one person that was so photo sensitive that we finally had to cut his contrast ratio down to two to one, 200 percent because anything higher hurt his eyes. Something that is probably not heard of, but people have sufficient variance in the way low vision affects them.

I hope that answered your question.

>> DAVID SLOAN: Wayne, could you repeat that piece of work you talked about? We missed the first part of what you said. You were muted at the time.

>> WAYNE DICK: Oh, the tool was Vickie Hanson and John Richards' tool. He's one of the reviewers. They invented a tool years ago that really started customization. It was really a fantastic tool. Microsoft withdrew the supporting technology for it. IBM didn't market it well. So we never really saw it. That was the name of the tool, Web Adapt To Me. And the initial papers on that are very interesting papers to read on text customization. I think Shawn references one of them in her paper.

>> DAVID SLOAN: Yes, I think that's correct.

Okay. Let me ask a few people who discuss how text customization requirements should be better addressed in accessibility guidelines and Web standards and other best practice guides? We have heard from Olaf on the potential benefits that designing for mobile in adopting a responsive design may have in supporting text readability.

Can I ask Anthony for your view on the aspect of selecting typography, whether it's the subtext rendered type faces or more specific typographical selection? To what extent would you like to see that implemented in guidelines or standards for Web content authors with application developers, user agent developers?

>> ANTHONY LEE: Well, I think obviously there is a balance to be struck there in the sense that if the user uses a style sheet and makes too many changes, it will send the website content into a bit of a tailspin. But from my perspective, in essentially a nontechnical perspective, first of all I would like it to be made easy to make changes to through an agent or whatever so that you can change type face, change things around the way it looks and feels.

As I said in an e-mail earlier, almost a bit like turning up the volume on a television. You just want it to be that easy to do that.

And at the moment I don't think it is particularly easy. There isn't very much of an understanding of how to do it, even by IT professionals. As I'm finding out myself.

Secondly, yes, I think for authors to be authoring content in a way that lends itself to being adapted through user sheets and there isn't too much of a gap between what the author, the way the author wants it to look and the way the user wants it to look, that would be enormously helpful.

One of the examples that was given is if you were an agent, you put it into high contrast and then lots of stuff just disappears off the website. If you know there's definitely a balance there to be struck, but most of all, I think the most important thing is for it to be really, really easy for the user to make those adjustments without having to talk to necessarily, talk to IT professionals. You can just do it through your graphical user interface. If that answers your question, that would be my summary.

The other thing I would say is I do think that mobile is becoming my best friend because sub-pixel rendering isn't deployed on mobile devices because of the way in which the screen has to rotate, first of all. And secondly, I've just started using a tablet. I am finding it easy to just zoom on a lot of things, to zoom HTML text, double tap it and everything reflows and it looks fantastic. That's easy to do without having to write special software. I'm just zooming and tapping. I think there is a lot that can be done to make things a lot better for people in the same boat as me.

>> DAVID SLOAN: Thank you very much, Anthony. Olaf has shared a comment on the IRC channel. Olaf, would you like to talk about that a little bit more? It's effectively a wish list item for Web content accessibility guidelines, Version 3 .

>> OLAF DRUMMER: A statement looking into the future and hoping for such a direction accordingly.

I think we should take advantage of what is happening anyway, driven by mobile content where lots of people think they can make tons of money, which is fair enough. And then just hijack that a little bit and say well, make that something that is also required from an accessibility point of view. If the Web content creators, if you can make your website work on the mobile devices as well as on the laptop, make it in a fashion that it works for accessibility needs. If they can easily surplant the styling that is provided with the Web page by my own styling, then I'm definitely gaining something. A lot of things that are difficult today, looking at the CSS style generator from Silas Brown show cases that. It would make things much easier. His agent would have to support that by making it easier for the regular user to load ar apply their own style sheet. At the moment it is really something for people who love technology, but not for the average user.

Maybe the whole hype around mobile can help us get there. That's it for my side.

>> DAVID SLOAN: Okay, thank you very much, Olaf. Something that occurred to me while I was listening to the recent discussions is a sense of frustration amongst some participants at the impact that, the negative impact that making a customization change has on the ability to effectively use a website. That made me wonder whether participants have comments on the role of evaluating websites for accessibility and the particular step of using different style sheets, making different customization in are order to see whether user experience is maintained to acceptable level.

Does anybody have any particular comments on the role of text customization as an evaluation technique? I realize this is possibly straying off topic, but I thought it would be interesting to get some comments.

Whitney, let me ask you to elaborate on the comment that you have just shared.

>> WHITNEY: What I keep seeing is that in usability testing with general audiences, people are often quite able to customize their screens the way they want them to be, personalize them with desktop backgrounds and so on. Yet necessary things that make it possible for people to read seem quite hard for them to be find. I wonder whether part of this is that we have isolated these features off from accessibility options that people don't relate to that word or think of that as a place to go or even think of themselves as technologically sophisticated enough to work with them.

So I want to adhere to all of the comments both in the discussion and on the chat session that suggest that we ought to be leveraging into the kind of general application, thinking more about universal design and something that says that everybody wants to customize in some way and we broaden the degree of customization possible there. That is not to say that research into how customization tools work isn't important because that's what helps us understand what customizations we need and how to present those customizations.

Do we want people to have to do something like select an exact line spacing or have something that says adapt this text here. Made it wider, tap tap tap. And then they say save it, I like my text like this. Something about the usability and discoverability of these features is as important as the features themselves.

>> DAVID SLOAN: I think that's a very interesting contribution in that it helps us to focus in on this balance between, if we are looking at the usability of customization support, perhaps we can argue that that includes the level of diagnostic support that Wayne was talking about and the role that smart systems in terms of detecting what might be appropriate for a particular user could be offered as a suggested default.

I would like to bring in Clayton Lewis if he's available. Yes, I hope he is. He just left an IRC comment. Clayton, you left a comment a short while ago which disappeared from my IRC log, discussing or wondering whether machine vision and content recognition could play a part in helping to determine structure and possibly applying an appropriate level of customization.

Could you elaborate further on this idea? And how you might see it moving forward?

>> CLAYTON LEWIS: Great. So I think one of the big background issues for all of this is the hold that visual design has on people who create content. Not everybody, but many people. And that cuts against what we are trying to do in a number of ways. One way is that people who are working towards those goals are working with tools in which they are not oriented towards underlying meaning and content, but they are really oriented towards presentation. They are implicitly expressing content in the way that Wayne and others have talked about, but they are not working on tools in which they make those things explicit. I would go beyond that and say that in many cases they don't want to do that. It would be difficult to accomplish a change in how they look at what they are doing because they think of their task as being one that is intrinsically quite visual in character.

Given that that is going to happen some of the time and I'm certainly not saying it is going to happen all of the time, but given that that will happen some of the time a possible avenue is something that retrieves the situation after it gets out of their hands. They've created something -- actually, I really like Wayne's example of the citation. Okay, one way to go there, you get the person to say that it's a citation and some of the time that will work and other times it won't.

When it doesn't, one can imagine a tool that actually recognize that is what is in this document at this point is a citation and recovers that semantic information that was not there before. In principle at least this approach could be applied to a number of the problems that have been mentioned along the way, as really being kind of practical difficulties in many situations such as the fact that when you translate the PDF into an HTML, the logical flow is lost. Well. if we could use recognition technology to determine what the logical flow actually is, then we could do better than we are able to do now. It really was a question and I wonder if anyone out there knows of work along these lines.

>> DAVID SLOAN: Does anybody have a positive answer to that? I personally am not aware of any. So at least it's something to note in the discussion in the symposium and possibly a forward research direction I also would like to reread Eileen's IRC comment: I wish the reader had a slider for the contrast similar to the slider for size.

This is an example of making customization simpler and more explicit to people in the way that -- I'm going to borrow Anthony's phrase one I've used in the past: To make it easier to apply different aspects of customization as it is to change the volume on a television set. I think that's an interesting analogy in terms of ease of use in customizing how information is presented to you.

I would like to ask Vasile if he would be able to share with us either by voice or by text a little bit more on his research looking at evaluating his text customization tool. If I can quote from your paper, Vasile, we realize the importance of detailed customization, and this part is under construction. The customization will address options like text size, text color, background color, font, character, line and paragraph spacing.

How will you determine what options to provide and what additional inputs might help you make that determination?

[Vasile was not connected to audio. He typed his comment and David read it.]
>>Vasile Topac: I try to rely on the existing literature with user studies about text customization. The guidelines I'm following for people with dyslexia are relying on the user studies provided by Luz Rello. So we've already heard from Luz today. The main customization will be focusing on font family, font size, background and text color, character word and line spacing. I collaborate with Luz on this and probably will have some user studies and tests done on text for all.

However, I'm lacking literature to find out about the layout customization since text customization can modify the original layout, is it better to try to preserve the original layout with a possible side effect on usability like the need for horizontal scrolling due to big tables? Or is it better to modify the layout to eliminate the need for horizontal scrolling but with the risk of affecting the logical structure?

>> DAVID SLOAN: And that to me very much reflects Gordon's earlier comments on the interaction of variables and the interplay or the effect of one on the other and seems to be a particularly challenging aspect of customization where it's very much a balance of getting a number of variables at an appropriate level that each contribute to an overall effect of improving rather than decreasing readability.

Does anybody have any specific comment in reaction to Vasile's feedback there?

Wayne, would you like to give your comment verbally?

>> WAYNE DICK: One of the things that Richards and Hanson did, they created a linearization algorithm for pages and it worked very well, he pages that meet the guidelines and are already set up so that there's a reading order for screen readers which can be linearized in that order. That will be consistent with a logical reading order of the page.

So I think in many cases reorganization is preferable because there's just no way that you can render a page, say, with a size that requires 25 words per page any other than a re-rendered format. There's a concept of the view size and the smallest size is one word at a time which is reading out loud.

And then the normal size is what normal people see. That's suitable for multiple column presentation. But once you get to a reasonably large print, single column really seems to be the only presentation that enables word wrapping. I think then that you really have to pay attention to ways of indexing around pages because say you're reading a mathematical article and you have to go back and look at a theorem. The ability to be able to go back and forth with hyperlink kind of performance is important.

So I think that basically what we have is a fundamentally different typography than print, but print translations should be possible. It's a different issue than text customization. It's like restructuring pages so that text customization can be applied.

>> DAVID SLOAN: Thank you, Wayne. Suzette, one of your comments, you briefly mentioned what we might learn from eBook readers and other devices in supporting easy reading, not with standing the differences in the stream design. Could you possibly share a brief reflection on what we might learn in terms of supporting effective customization from an eBook reader device?

>> SUZETTE: Well, certainly I have been really surprised by what happened with the eBook readers. They seem to have gone against some very established rules we have about how to lay out the text on the screen because they have gone across to using a serif face rather than a sans serif face. And generally trying to create something that resembles something much more of an paper ground. It's an off-white background.

My challenge I was trying to get at is that we have a whole load of recommendations really based on best practice at the time they were written. I really think that the current technologies like the eBook reader are actually maybe we should be rethinking some of our recommendations because the quality of the image has just gone and made such huge progress. And the mobiles as well, of course.

>> DAVID SLOAN: Okay. You're suggest ugh that in the similar way that we can look to inspiration to how design approaches have adapted to optimized mobile experience, perhaps we can do something similar to learn from the positive changes that have been implemented by eBook readers design and display of content.

We are heading close to the end of the time we have available. There is one question from Anthony that I would quickly like to ask him to share with us.

Would you like to go ahead?

>> ANTHONY LEE: So yes, it was, I think again going back to the sort of way in which mobile seems to be helping, one thing I have noted again from the user perspective is when I'm looking at my hot mail account through the app, on a tablet, some e-mails come in and I zoom it to be able to see it easily. And -- or more easily. As I say, I double tap and it all reformats and flows, et cetera. Some stuff doesn't. That's just a frustration. That must have something to do about the way in which the content, the e-mail has been authored. Again that sort of goes back to my point about if there are some sensible standards that are implemented that will allow that sort of thing to happen, that would be quite a leap forward. It's simple to a use, but it seems impossible for the IT industry to achieve consistently.

>> DAVID SLOAN: It certainly sounds to me like the importance of e-mail clients as Web content authoring tools is something that is still under-recognized and something we need to deal with.


>> DAVID SLOAN: Thank you, Anthony.

I would like to ask Shawn to contribute before she steps in as Co-chair to finish off the symposium. So Shawn, as an author would you like to comment?

>> SHAWN HENRY: Just to answer the question you asked about standards and guidelines, and also just to reflect on where the discussion has gone today, I do think that text customization functionality, is a accessibility requirement and if a user agent does not provide that functionality, it does not sufficiently support accessibility. I think it is interesting, the discussion -- we ended up talking a lot about user style sheets and I think that one of the things to look at is advanced users currently can do a lot with user style sheets, with most Web browsers.

But that text customization function is not available for all agents, even for advanced users. One of the things that I just wanted to make clear, I think a lot of people have alluded to is that in addition to the advanced, the functionality for advanced users we should have some simple basic customization available for less advanced users.

And then the final point is, while a lot of text customization functionality resides with the user agent, I think we have made it clear in the discussion today and with the contributions in voice and e-mail and chat, that there is an important role for content, that Web content does have some responsibility to ensure that when users customize text that their website is still usable.

So that was good for my author hat. Back to you, Dave. You wrap up and then I will.

[Next steps - Symposium Report, Invitation for additional short contributions]

>> DAVID SLOAN: We are just about out of time. While there were a couple more questions in the IRC chat we should really wrap up now.

As Shawn I think will tell you, the conversation has not closed. It will continue. I think we've identified a number of different research questions from the usability through to exploring new approaches to gathering information about specific needs of specific groups in order to push forward text customization as a really important part of the Web accessibility jigsaw.

Thank you very much for your contributions. I'm now going to hand back to Shawn to close the symposium. Shawn, over to you.

>> SHAWN HENRY: Thank you, Dave. I just wanted to remind everyone that the next step is for us to work on a research report. So there are additional opportunities for you to contribute for this topic and for the research report.

There is an e-mail list that is publicly archived. [(public-wai-rd-comments@w3.org) mailto RDWG Comments list, with subject starting "[TC4R Symposium]"] We encourage you to sendany information to that list that would further discussions on this topic. Feel free to send links to related research or personal observations or any additional information that you think might contribute to this topic, and particularly for the research report.

The text transcript of the discussion today will be available online, as will the chat. So there will be additional opportunity to share this information and we would just ask that if you are sending comments, if you would ideally do it soon or certainly within the month of November, that would be good.

I put in the chat the e-mail address. It is also in the symposium page. If you could preface your comments with the acronym for the symposium it will be a little easier to find, but that is not a requirement.

We thank you very much. Again thanks to the Working Group and to the Scientific Committee and especially to the paper authors, for everyone who sent additional information or who will be sending additional information. For those who contributed on the phone, we appreciate all your discussion and input. We look forward to sharing more information through the e-mail list and through the draft research report which will be available for everyone to review.

Thank you very much, Dave Sloan and Shadi Abou-Zahra and everyone who was involved in making this a successful symposium.