Re: IBM's blindenfreundlicher Web Browser

[Peter Brass <p.brass_bEi_snafu.de>]

At 21:42 03.02.99 -0500, you wrote: 
>
> >Fuer eine Demoversion, natuerlich in englisch, schaut mal unter 
> ><http://www.ibm.com/sns>www.ibm.com/sns. 



die Demoversion ist dort z. Z. noch nicht verfuegbar, ich habe aber mal ein
Referat runtergeladen, das auf der "Closing the Gap" Konferenz im Herbst des
vergangenen Jahres vom Entwickler gehalten wurde. Es ist in Englisch, aber wer
sich der Muehe unterziehen will, kriegt ein paar gute Eindruecke, was das
Programm denn mal koennen soll.

Viel Spass

Peter
----------------cut here-----------
Home Page Reader: IBM's Talking Web Browser

Chieko Asakawa 
IBM Japan Ltd., Tokyo Research Laboratory 
chie_bEi_trl.ibm.co.jp 

Catherine Laws 
IBM Special Needs Systems, Austin, Texas 
claws_bEi_us.ibm.com 
 



Problem Statement

With the World Wide Web attracting students, employees, and families,
including
those with disabilities such as visual impairments, the need for accessible
web
environments will soon become an essential part of life. However, people who
are blind cannot access graphical web data or text presented in complex
tables,
forms, and frames unless it is presented in a logical, structured format along
with speech and/or Braille output and keyboard-accessible features (Brewer,
Dardailler, & Vanderheiden, 1998). Unless web page authors become aware of and
familiar with accessibility problems and implement the latest strategies and
techniques for overcoming those problems, web pages will remain inaccessible. 

Even if web authors developed accessible web pages by following accessibility
guidelines and by using validation tools to correct accessibility problems,
the
web could still be inaccessible for visually-impaired students and adults
if an
accessible web browser solution is not available. Blind users often rely on
dependent talking web browser solutions, which include commercial graphical or
text-based browsers that depend on and cooperate with screen readers, to
provide speech output and web page navigation. Recently emerging as
alternative
solutions for blind and visually-impaired users are independent talking web
browsers, which are self-voicing browsers that provide speech output and web
page navigation based on direct interpretation of HTML and other web page
constructs (Gunderson & Jacobs, 1998). 



However, talking web browser solutions suffer from a growing list of problems
and critical needs (Vanderheiden, Chisholm, & Ewers, 1996). The problems
include the inability for a user to easily obtain page layout information,
difficulty navigating the grammatical and structural elements of a web page,
synthesizer inaccuracies, the inability to accurately read tabular
information,
and limited document search capabilities. The list of some of the features and
strategies that browser solutions need to incorporate to help alleviate these
problems include: 

•Easy navigation between the main browser window and the URL input window.
•Table cell reading along with row and column headings for each cell.
•Navigation by grammatical and structural page elements, such as headings,
paragraphs, lines, words, characters, hypertext links, lists, maps, tables,
forms, and frames. •Full document search capabilities. 

In the last few months, the W3C, as part of its Web Accessibility Initiative,
has authored their first working draft version of accessibility guidelines for
user agents, which include screen readers and screen magnifiers working with
web browsers as well as independent web browsers (Gunderson & Jacobs, 1998).
These prioritized guidelines offer recommendations to browser manufacturers
which significantly improve access to WWW documents. The guidelines discuss: 

• Control over browser defaults and author styles. • Alternative
representations and control for images, maps, multimedia, embedded
applications, and tables. • Alternative document views. • Maintenance of
document view and focus. • Document summary information. • Element and event
identification. • Sequential and hierarchical navigation within a view. •
Navigation for views, pages, tables, menu commands, and elements with DHTML
events. • Direct access navigation. • Accessible configuration and
documentation of accessibility features. • Cascading Style Sheet support. •
Compatibility with HTML 4.0 and with third-party assistive technology. 

This paper discusses how IBM Home Page Reader, a new talking web browser
solution, implements many of these guidelines in its approach to address World
Wide Web accessibility problems for blind users. 
 



History

In 1996, blind people in Japan had only two sources of published information:
Braille books and cassette tapes. With computer users able to get information
easily and quickly from all over the world using the Internet and blind users
unable to access the web easily, the information gap between sighted and blind
users was becoming wider. When Japanese blind users did try to access the web
in the DOS environment, the screen reading and navigation of hypertext links
and two-dimensional information was difficult and incorrect (Asakawa & Itoh,
1997). 

Since the IBM Screen Reader/2 product had been translated into Japanese, the
IBM Tokyo Research lab first tried to create a prototype system using SRD/2 to
read Netscape Navigator web pages. However, this solution read only the text
information displayed on the screen; it was unable to read and navigate
tables,
forms, and frames. In addition, users had to manually scroll through web pages
that were larger than one screen in order to read through an entire web page.
To address these problems, the research lab decided to develop a talking web
browser solution for Japan that analyzed HTML tags rather than simply reading
the screen. In October, 1997, the IBM Japan Entry Systems Business Unit (ESBU)
announced IBM Home Page Reader as a Japanese consumer product for blind users
to access the web. This year, the IBM Special Needs Systems organization in
Austin, Texas, has worked with IBM Japan to develop Home Page Reader as a U.S.
English, IBM Independence Series product that offers blind users better access
to the World Wide Web. 
 



Overview

IBM Home Page Reader offers a number of features that enable it to provide
better access for blind users to the World Wide Web: 

•Netscape Navigator synchronization. Because Home Page Reader communicates
with
Netscape Navigator to get web information (HTML source), HPR reads tables,
frames, forms, and images on graphical web pages logically, as they should be
read. While HPR speaks web pages to a blind user, sighted users can see the
Netscape rendering of the same web page the blind user is reading to provide
sighted assistance, if needed. •Software text-to-speech support. HPR supports
IBM's software text-to-speech, SAPI-compliant engine, IBM ViaVoice OutLoud. No
extra hardware expense is required. •Numeric keypad web page navigation. HPR
provides a logical, one-handed numeric keypad interface for navigation and
manipulation of web page elements, such as image and text links, text, form
elements, tables, maps, lists, headers, and frames. Keys help mode allows
users
to press any HPR key combination and hear its description. •Web pages are read
with speed and auditory distinctions. HPR's fast-forward key allows users to
skim web pages to locate desired information quickly. Using a female voice to
read links and a male voice to read text, HPR provides the user with easy,
auditory distinctions when reading web pages. Optionally, the user can select
the word "link" or a MIDI sound to identify links, or select no
announcement of
links, if desired. •Web page authoring accessibility features are utilized.
HPR
reads ALT or other descriptive text, when it exists, for all images, maps, and
other web page objects. Otherwise, it reads the URL link information. HPR
reads
other new HTML 4.0 information provided by web authors, such as table captions
and summaries. •Web page orientation. "Page summary" information and "where am
I" commands tell the user information about the number and location of page
elements on the current web page and at the current location. •Online help and
bookmarks access as web pages. HPR online help and bookmarks are easily
retrieved as local web pages with hypertext and help the user get started
quickly using HPR. •Electronic mail feature. HPR provides an integrated
electronic mail feature consisting of forms and menus that can be read and
handled by HPR. This feature requires the Microsoft Personal Web Server or
Peer
Web Services. Also, for sending mail, HPR can handle mailto tags. •Easy,
self-voicing installation. Spoken instructions guide non-sighted users through
a standard installation program. HPR setup includes installation of the IBM
ViaVoice OutLoud software TTS engine and an optional installation of Netscape
Navigator for new Internet users. 



HPR Feature Implementations



Web page reading and navigation



Logical numeric keypad layout

HPR uses a logical layout of the numeric keypad for its users to read and
navigate web pages. Basic numeric keypad keys are assigned to read the
previous, current, and next link (1, 2, and 3), page element (4, 5, and 6),
and
word or character (7, 8, and 9), to read the page (0) and to stop reading
(Enter). Other basic keys provide access to a history list (Num Lock), online
help (slash), a settings menu (asterisk), and bookmarks (minus). In settings
mode, basic keys (2, 4, 6, and 8) are used to logically navigate and select
settings in a non-visual menu system. 

Extended HPR functions use the plus (+) key then a basic key that is usually
related to the extended function. Extended numeric keypad keys are assigned to
read the first or last link (+ then 1 or 3), page element (+ then 4 or 6), and
word or character of an element (+ then 7 or 9). Other extended functions
include reloading a web page (+ then Num Lock), getting keys help (+ then
slash), adding or deleting bookmarks (+ then -), setting word/character
reading
mode (+ then 8), getting "where am I" information (+ then 5), opening a
link (+
then 2), fast-forward page reading (+ then 0), getting a new URL or searching
(+ then dot), and cancelling a connection (+ then Enter). 



Jump keys in Home Page Reader use the dot (.) key to provide additional
navigation capabilities for tables (dot then Num Lock, slash, or *) , headers
(dot then 1, 2, or 3), frames (dot then 0), paging up and down (dot then 4 or
6), and structures (dot then 7, 8, or 9). A structure can be a list, select
menu, table row, form, or map. A page summary key (dot then 5) is also a jump
key. 



Functions for managing bookmarks require the user to hold down the bookmarks
key (minus) and then press a basic key. Functions for history list navigation
involve holding down the history list key (Num Lock) and then pressing a basic
key. 



Editing and clipboard functions combine basic HPR keys with standard Windows
keys to provide standard screen reader character echo and browsing
capabilities
in text input fields. 



The following sections describe how HPR keys address many of the current W3C
WAI accessibility guidelines for user agents (Gunderson & Jacobs, 1998). 



Sequential navigation

HPR provides keyboard commands to move sequentially between all types of links
using the previous, current, and next link keys (1, 2, and 3). Links include
all clickable HTML hypertext and controls, such as text entry fields,
pushbuttons, radio buttons, check boxes, selection menu items, images, image
map areas, audio files, and frames. When reading web pages, HPR identifies
links with a female voice and normal text with a male voice. Optionally, the
user can select to identify links with the word "link" or a MIDI ping, or with
no announcement. 

HPR also provides keyboard commands to sequentially navigate all page elements
or paragraphs (using the 4, 5, and 6 keys) and words or characters (using the
7, 8, and 9 keys). 



If the user wants to navigate more quickly through a web page by finding
structured elements like maps, select menus, forms, lists, and table rows, HPR
provides jump previous, current, and next character keys (dot then 7, 8, and
9). When HPR automatically reads the page when first loaded or when the user
presses the play key (0), some of these structures, such as maps and select
menus, are announced along with their number of elements, but each element is
not read. To read each element in these structures, the user can find the
structures with the jump character keys and then read each element with the
item keys (4, 5, and 6) or link keys (1, 2, and 3). . 



If the user wants to "scroll" through a web page, they can press the jump
previous or next item keys (dot then 4 or 6) to move up or down the web
page by
10 items. 



If the user wants to quickly browse a web page, they can press the extended
key
(+) and then hold the play key (0). This command key makes the reading of the
web page sound like a cassette tape recorder that is in fast-forward mode. 



View navigation

HPR can read and open the views for all types of links when a link is selected
and the user presses the extended current link key (+ then 2 or double-click
2). When the extended current link key is pressed and a text entry field is
selected, the view becomes the HPR input dialog. For a radio button, the
button
state changes to on and all other radio buttons in the group are turned off.
For check boxes, the box becomes checked. For a single select menu item, the
menu item is selected and all other menu items are unselected. For image
links,
image map areas, and hypertext links, the referenced web page is loaded and
HPR
begins to speak the new page. For a mailto tag, HPR opens a mailto form. For
multimedia files and Netscape plugins like RealAudio, HPR starts the
multimedia
or plugin. 

When HPR loads a web page with a frameset, the view becomes a list of links
identifying the frames on that page. The user can open each frame as a new
page
by selecting a frame link and pressing the extended current link key (+ then
2). To return to the frameset, the user presses the jump play key (dot then
0).




Page navigation

HPR provides an initial bookmarks page that contains links to web pages in the
categories of Accessibility, News and Sports, Search Engines, Weather and
Travel, Commerce, Computing, and Home Page Reader Support. The user can visit
the bookmarks page by pressing the bookmarks key (minus) and can visit any
document on the bookmarks page by pressing the extended current link key (+
then 2). 

To browse a history of visited pages, the user can hold down the history key
(Num Lock) and press the slash and asterisk keys to move up and down the
history list. When the history key is released, the selected page in the
history list is loaded and read by HPR. 



To reload the current page, the user can press the extended history key (+
then
Num Lock). 



To get a new web page or a local file, the user can press the extended jump
key
(+ then dot) and type the new URL or filename. 



Direct access navigation

To move directly to page elements, links, and controls on a web page, the user
can search for its text or ALT text by pressing the extended jump key (+ then
dot) to display the input dialog. Then the user types "/" followed by the text
or ALT text and presses Enter. To find successive occurrences of that search
string, the user holds down the stop key (Enter) and presses the previous or
next item keys (4 or 6). 

The user can also move directly to the first or last item on a page by
pressing
the extended previous or next item keys (+ then 4 or 6) and to the first and
last link on a page by pressing the extended previous or next link keys (+
then
1 or 3). 



Hierarchical navigation

For web pages that use headings to delineate sections, the user can press the
jump previous, current, or next link keys (dot then 1, 2, or 3) to navigate
the
headings. This key is helpful for navigating the HPR online help pages,
research papers, and other online documentation. 

Table navigation

In most documentation, tables are used for organizing and reporting data.
However, many web page authors use tables for formatting web page elements.
Since HPR cannot easily identify which tables are used for formatting, table
navigation is difficult. To help the user navigate tables and obtain table
cell
information, HPR provides table navigation keys (dot then Num Lock, slash, or
asterisk), table row navigation keys (part of structure navigation keys with
dot then 7, 8, or 9), and header information provided when reading cells (with
1, 2, 3 and 4, 5, 6 keys) and obtaining "where am I" information (+ then 5).
When in a table, the "where am I" information also announces table captions
and
summaries. 

Other HTML 4.0 authoring accessibility features

HPR renders and reads the ALT text, title, or long description for images,
image map areas, and objects such as applets. If ALT text is blank, HPR
ignores
the image, area, or object. If no ALT text, title, or long description exists
but the image or area is a link, HPR reads some of the URL information. 

For server side maps, HPR identifies them as server side maps with the
coordinates (0,0). If the user clicks on a server side map, the coordinates
(0,0) will be sent back to the server and an alternative form of the server
side map may be returned for HPR to render. 



Web page orientation

To find out the title and the number of links, forms, frames, tables, and page
elements on the current web page, the user can press the jump current item key
(dot then 5). 

To obtain "where am I " information, the user presses the extended current
item
key (+ then 5). The "where am I" information consists of an announcement of
the
current structural item (map, menu, form, table) and a relative position
within
that item, as well as an announcement of a position relative to the whole
page,
such as the element or link number and the total number of elements or links.
Additional information may also be announced depending on the type of
structure, such as the type of a list or a control as well as table headers
and
summaries. 



Online help and keys help

Online help consists of local hyperlinked web pages that describe all the HPR
keys and their functions and other user interface considerations. These web
pages are opened by pressing the help key (slash). 

To learn or confirm the key assignments for HPR commands, the user presses the
extended help key (+ then slash). In keys help mode, when the user presses a
key or key combination, the system speaks the associated name of the HPR
command and its function. 



Configuration features

To configure HPR features, the user can press the settings key (*) to enter a
non-visual menu system. In the menu, the user presses the 2, 4, 6, and 8 keys
to navigate the settings menu and change the values for settings. Settings
include the speaking rate and volume, links announcement, fast forward
reading,
the startup page, various history list settings, and the Netscape
synchronization. 

With sighted assistance or a screen reader, users can remap HPR keys to other
keys on the keyboard. For using HPR with a laptop, a separate numeric
keypad is
recommended. 



Screen Reader compatibility

Since HPR requires a screen reader for access to some of its advance
customization features, it provides some screen reader compatibility, such as
the use of standard Windows controls. However, HPR must be silenced when using
a screen reader and vice versa. The Ctrl+F12 key silences and reactivates HPR
keys and functions. Unchecking a setting called "Active in the background"
also
enables HPR to coexist with a user's screen reader. The online help provides
additional information that suggests how specific screen readers can coexist
with HPR. 

How to obtain HPR

For marketing information about IBM Home Page Reader and other IBM
Independence
Series products, visit the IBM Special Needs web site at
<http://www.ibm.com/sns>http://www.ibm.com/sns. In addition to the web
site, you
can obtain information about IBM SNS products by calling 1-800-IBM-CALL and
through our dealers, such as MicroAge. 
 

References

Asakawa, C. and Itoh, T. (1998). User interface of a Home Page Reader.
Proceedings of the ASSETS '98 ACM Conference on Assistive Technologies,
1998-4.


Brewer, J., Dardailler, D., and Vanderheiden, G. (1998, February 16). Toolkit
for promoting web accessibility [Online].
<http://www.dinf.org/csun_98/csun98_057.htm>http://www.dinf.org/csun_98/csu
n98_057.htm 



Gunderson, J. and Jacobs, I. (1998, July 3). WAI accessibility guidelines:
User
agent. Working draft. [Online].
<http://www.w3.org/TR/WD-WAI-USERAGENT>http://www.w3.org/TR/WD-WAI-USERAGENT/ 



Vanderheiden, G., Chisholm, W., and Ewers, N. (1996, March 27). Making screen
readers work more effectively on the web. [Online].
<http://trace.wisc.edu/docs/screen_readers/screen.htm>http://trace.wisc.edu
/docs/screen_readers/screen.htm 

*******************************************************************
Happiness is something that comes into our lives through doors we
don't even remember leaving open. -Rose Lane