             COMPUTER SCIENCE UPDATE - SUMMER, 1995


                          Published By
                                
                National Federation of the Blind
                       in Computer Science
                                
                     20 Northeast 2nd Street
                          Apartment 908
               Minneapolis, Minnesota  55413-2265

                     Phone:  (612) 379-3493
                 Internet: curtisc@winternet.com
                  ELECTRONIC INFORMATION KIOSKS
                   WHAT'S ALL THE FUSS ABOUT?
                         by Curtis Chong

If you attended the 1994 meeting of the National Federation of the
Blind in Computer Science, held in Detroit, Michigan, you may
recall the discussion that took place about electronic information
kiosks.  Essentially, these are public terminals equipped with
touch screens and multimedia (voice and video) output, which are
supposed to provide greater "accessibility" to facilities,
information, and services.  Many of us were struck by the irony in
the use of the term "accessibility" in the video presentation we
saw; for as it turns out, these devices are, for the blind,
anything but.

Is this lack of accessibility something for blind people to get
excited about?  Will our lives be irrevocably damaged by the
proliferation of electronic information kiosks throughout the
country?  What is an electronic information kiosk anyway?

Generally speaking, an information kiosk is a public computer
terminal designed to give and receive information.  Information is
usually presented on a video display, accompanied by some
pre-recorded digitized speech.  Some kiosks contain a printer that
can be used to produce selected hard copy.  There is typically no
QWERTY keyboard or numeric keypad.  Information is entered through
virtual push buttons, which are displayed visually on the screen. 
For more complex data entry, a virtual numeric keypad or full
QWERTY keyboard can be displayed.  Users then touch the keys they
wish to "press."  Information presented on the screen is displayed
using a combination of text and still or animated pictures.  The
virtual push buttons can be located anywhere on the screen, and
there are typically no buttons in standard locations that can be
located by touch.

The technology represented by electronic information kiosks is
spreading rapidly across the country with little if any thought
given to whether or not it is useful to persons who are blind.  At
least, this is the way it appears today.  As I see it, we are faced
with two fundamental questions:

1.   Will the proliferation of electronic information kiosks--at
     least in their current form--harm our prospects to compete on
     terms of equality with the sighted?  To put it another way,
     will the widespread use of this technology force blind people
     to rely upon sighted assistance to perform those basic tasks
     which we perform independently today?

2.   If we wish to have full and equal access to the services
     available through information kiosks, what solutions should we
     be pushing for?  In other words, how do we want to interact
     with an electronic information kiosk?  Do we want to deal with
     it on its own terms (i.e., using the touch screen instead of
     a regular keyboard) or do we want to push only for those
     modifications that would make it possible for us to use
     information kiosks with our own access technology (e.g.,
     personal computer, Braille 'n Speak, or whatever.)?

Although I do not regard the proliferation of information kiosks as
an immediate threat, I do believe that we should begin formulating
approaches to adapting the technology so that it can be used
without sight.  State and federal agencies find kiosks appealing
not only because they give the appearance of increasing public
access to the government, but also because, in the long run, they
represent a way to reduce staff while providing the same amount of
service.  We are already seeing kiosks spring up in California,
Washington, Texas, and other states.  The White House has announced
the opening of a Worldwide Web Server that will be accessed via
electronic information kiosk.  In my home state of Minnesota, the
Department of Economic Security, the department which houses our
state agency for the blind, is seriously considering the use of
information kiosks as a way to expand the reach of its services. 
Kiosks are simply too appealing to resist.  So, although today it
is no big deal for us not to have access to government information
and services through these kiosks (we can always visit an office
where there are live human beings to provide the service), what
will happen if the decision is made to close down a local office
because the service can be provided more cheaply and efficiently
through an electronic kiosk?  Is it a stretch of the imagination
for a bank or airport to consider replacing live staff with an
automated kiosk?  I don't think so.  The day will come when
electronic kiosks are as common as automatic teller machines, and
it is incumbent upon us as knowledgeable and concerned blind
consumers to have a say about how we will interact with this
technology.

Turning to the second question--how will we access electronic
kiosks--I will be the first to admit that we do not have all of the
answers.  We don't have enough experience with touch screens and
the like.  We certainly aren't lacking for opinions, however.  Some
people think that all electronic kiosks should be equipped with a
telephone handset that will put you in touch with a live operator. 
Others maintain that every kiosk should have a real keyboard that
can be operated by touch and a speech synthesizer with
intelligently designed software to drive it.  A third viewpoint
holds that kiosks should be equipped with inexpensive, infrared
transmitters designed to send standard signals to  whatever access
technology the blind individual wants to use, thereby permitting
output to be delivered in the preferred medium.

Steve Jacobson (NFBCS Vice President) and I had an opportunity to
examine two different approaches.  While attending a technology
conference in Bloomington, Minnesota, we both examined a prototype
talking kiosk application put together by Dr. Gregg Vanderheiden of
the Trace Research and Development Center in Madison, Wisconsin. 
Dr. Vanderheiden took the source code from an existing kiosk
application running on the University of Wisconsin campus and
modified it to work on a PC equipped with a touch screen and an
Accent speech synthesizer.  Here is how it works.  Using a pen or
the back of the fingernail, one can slide across the screen and
hear each item as it is traversed.  If one wants to select a
particular item, one simply lifts the hand off the screen, and the
item is "selected."  To avoid selecting anything, one simply moves
the hand to some white space, denoted by a hissing sound generated
by the synthesizer, and lifts.  In addition to buttons that are
unique to each screen, each layout contains a few "standard"
buttons.  For example, there is one button to describe the screen
and another to take you to the "Main Menu."  By selecting the map
of the college campus, it is relatively easy to locate the
restrooms and other features of a building.  One can even search
the student directory by entering entries via a virtual QWERTY
keyboard, and a security code can be entered from a virtual numeric
keypad displayed on the screen.  Initially, proficiency suffers
somewhat, but with a bit of practice, one can develop enough speed
to use the technology with good efficiency.

Dr. Vanderheiden believes that research still needs to be done to
determine the best way to design what he refers to as a "seamless
interface."  There are some questions that need to be answered:

o    What is the most comfortable way for a blind person to
     interact with a touch screen?

o    If talking technology is a standard feature on information
     kiosks, should speech be generated for everybody or only for
     the blind user?

o    If only for the blind user, what standard sequence should be
     developed to activate the talking interface?

The second approach Steve and I examined was incorporated into a
demonstration kiosk developed by North Communications.  Upon first
examination, it becomes clear that this kiosk is not designed
initially to be accessible to the blind; it is designed to provide
information about the services offered by the Minnesota Department
of Economic Security as a whole.  The only concession made for
purposes of accessibility have to do with the information segment
on services to the blind.  Every three or four minutes, the kiosk
will generate some music and a verbal prompt instructing the user
to touch the top part of the screen to hear about services for the
blind.  If the user responds accordingly, various information
segments can be selected by pressing one of the four corners of the
screen.  (Needless to say, we were quick to inform North
Communications of our desire to have all of the information
segments accessible--not just those dealing with services for the
blind.)

The kiosk contains design features which are clearly meant to be
used by someone who can see the screen.  For example, for many menu
selections, users are instructed to "touch here," meaning that they
should press a button highlighted by a moving arrow.  Steve and I
suggested to North Communications that this verbal prompt be
changed to "touch the screen now," and the software modified to
react positively when the screen is touched within a few seconds of
the prompt.  We also pointed out that additional buttons could be
added with no change to the physical hardware.  For example, it is
fairly simple to locate nine positions on the screen by touch: the
four corners, the centers of each edge, and the center of the
screen itself.  It requires only a software change to prompt the
user to touch one of these nine locations to make a menu selection.

A lot of the information in the North Communications prototype
consists of pre-recorded verbal presentations.  However, there is
one application (a job selection screen) which displays information
about a variety of jobs in plain text on the screen.  This
application is totally inaccessible to the blind.  It can only
become accessible by incorporating new technology into the kiosk
(e.g., synthetic speech or some sort of infrared transmitter).  It
is interesting to note that the software driving the kiosk is
running under IBM's OS/2 Presentation Manager.  We know that OS/2
applications can be made accessible to the blind with IBM's Screen
Reader/2.

The North Communications kiosk is not a touch screen per se. 
Compared to the touch screen in Dr. Vanderheiden's prototype, the
North Communications kiosk is sluggish, recognizing button presses
only once every few seconds.  The technology is not designed to
accommodate scanning with a pen or fingernail.  On the plus side,
the kiosk does generate a beep when a button is pressed.

In discussing ways to make the North Communications kiosk more user
friendly to the blind, we found that many of the applications could
be made useable by modifying them to incorporate the "touch now"
concept, discussed earlier.  Many speech prompts can be built into
the kiosk by using pre-recorded phrases.  But for those
applications that convey complex and often changing information, it
is clear that pre-recorded speech simply is not the ultimate
answer.

Based on what you have read so far, you might conclude that we
should push vigorously to have Dr. Vanderheiden's approach
incorporated into all new kiosks.  However, the answer is not quite
that simple.  The Vanderheiden prototype was examined by blind
people who are admittedly more sophisticated in their use of
computer technology than the average person who is likely to walk
up to and use a kiosk.  This "average" user is assumed not to be
familiar with computer technology--in other words, to be "computer
illiterate."  In general, information kiosks are meant to be used
by such individuals.  Which approach would work well for a
similarly situated blind person: the Vanderheiden prototype or an
appropriately modified North Communications kiosk, equipped with
synthetic speech or other output capabilities?  The answer is, at
best, inconclusive.  The one clear fact is that we, blind
consumers, need to participate actively and affirmatively in the
formulation of any solution that is developed on our behalf.  We
are fortunate in that Dr. Vanderheiden's prototype will be
available for us to examine in detail at the 1995 National
Federation of the Blind Convention.                    NFBTRANS VASTLY IMPROVED
                         by Curtis Chong

On August 13, 1992, David Andrews, Director of the International
Braille and Technology Center, announced that the National
Federation of the Blind was releasing the source code for NFBTRANS,
its well known Grade II braille translation software.  "We are
releasing the source code to the computer community as a gift, not
to enrich individuals or companies," Andrews said.

Randy Formenti, an enterprising computer programmer who happens to
be blind, took the PASCAL source code, converted it to C, and
produced several new and improved versions of the program, which he
then uploaded to NFB-NET, the National Federation of the Blind's
electronic bulletin board.  According to Formenti, "My purpose in
making improvements to NFBTRANS is to make a high quality braille
translator available free to anyone who has a need for it and to
promote braille literacy."

The latest version of NFBTRANS now available from NFB-NET is
Version 7.39.  Here is a list of the improvements that have been
incorporated into the program by Randy Formenti:

1.   The user now receives meaningful error messages instead of
     cryptic run time errors with no explanation.

2.   The nfbtable program is no longer needed.  The table of
     translation rules is now read from a text file that can be
     edited by the user.

3.   NFBTRANS now has a configuration file.  The user can select
     from over fifty options which can change the behavior of the
     program.  For example, if you always want to emboss a single
     copy and start on page one, you can set this up in a
     configuration file.  If this is the default configuration
     file, NFBTRANS will no longer prompt you for this information. 
     In fact, you can set things up in the configuration file so
     that NFBTRANS only asks you for the name of the file to be
     processed.  No more annoying and repetitive menu prompts.

4.   Enhanced command line options permit virtually any default
     value to be temporarily overridden.  In addition, one can
     specify alternate configuration files, adding even more
     flexibility to the program.  A command line option can direct
     NFBTRANS to translate a list of files whose names have been
     stored in a special list file.

5.   NFBTRANS can be instructed to process groups of files
     differently, depending upon the file extension.  For example,
     suppose you store all of your C programs in files using the
     ".C" extension.  NFBTRANS can be configured to always emboss
     all files with the ".C" extension using eight-dot braille.

6.   Interpoint braille is supported.  Page numbering of even (left
     facing) pages is suppressed, and provision is made to skip to
     the next right-facing page.

7.   Automatic table of contents generation with braille page
     numbers has been implemented.

8.   You can choose how NFBTRANS will detect a paragraph in the
     original text: block style, indented lines, or both.

9.   Several new commands for table formatting have been added.

10.  A conditional end-of-page command has been added to prevent
     titles from being separated from subsequent text by page
     breaks.

11.  Several commands have been added to the external format
     language to permit automatic alteration or deletion of words
     or lines containing specific character strings.

12.  The user can now control whether NFBTRANS produces audible
     tones and whether page numbers are displayed to show progress
     of translation.

13.  Commands are now provided that give the user control over how
     long, centered lines are broken and the number of spaces the
     carry-over list lines are indented.

14.  Files can be embossed as they are being translated or in the
     backround using the DOS print command.

15.  NFBTRANS can be configured to output a statistics file
     containing items such as estimated embossing time and words
     per page.

16.  NFBTRANS contains support for foreign language braille
     translation tables and math translation tables.

NFBTRANS comes with complete documentation and C source.  It will
compile under MSDOS or Unix.  NFBTRANS is available free from
NFB-NET or the handicap ftp site.  For a uuencoded file send a
request to Randy Formenti.  His Internet address is rcf@genrad.com.                             MINUTES
                         ANNUAL MEETING
      NATIONAL FEDERATION OF THE BLIND IN COMPUTER SCIENCE
                          JULY 3, 1994
                   by Mike Freeman, Secretary

The 1994 meeting of the National Federation of the Blind in
Computer Science (NFBCS) was held at the Westin Renaissance Center,
Detroit, Michigan, on the afternoon of Sunday, July 3, 1994 during
the convention of the National Federation of the Blind.  President
Curtis Chong called the meeting to order at 1:01 p.m.  He announced
that the NFBCS newsletter would be available on NFB-NET and in
Library 5 of CompuServe's Disabilities Forum.  Brian Buhrow will
upload the newsletter onto the Internet he said.

President Chong next discussed the new computer, the Myna, produced
by Technologies for Independence (TFI).  He said that TFI has been
very friendly toward the NFB. Describing the Myna as "interesting"
and "different", President Chong urged his listeners to look at the
Myna and tell TFI how it could be improved.  He said that although
it is not an ideal system, it can be made so through constructive
suggestions.  He said that it does things other "palmtops" don't
do; it is the smallest computer with DECTalk in it.  He then
introduced Mohymen Saddeek from TFI to describe the Myna.

Mr. Saddeek described the Myna as weighing about 1.2 pounds and
having dimensions of 9.6 inches by 4.6 inches by 1.16 inches.  It
is a 386 computer with an IBM-COMPATIBLE keyboard comprising 82
keys with imbedded numeric and function keys.  It has DECTALK
built-in, 5th-generation chips and software on flash ROMS.  It
comes with the MicroSoft Works suite of programs and a full
implementation of IBM Screen Reader for DOS.  There is, however, no
need to use an extended keypad. Profiles are included for
WordPerfect, Lotus, MicroSoft Works and Microsoft Organizer.  The
Bookmanager program (for reading RFB E-Text books) is available. 
The Myna has two memory expansion slots; memory can be expanded to
40 megabytes.  One can buy the Myna with a "doubler" for a possible
memory capacity of 80 megabytes.  The built-in software can be
updated.  The introductory price is $1,995 with AC adapter,
carrying case and headphone.  Two braille displays will be
available in September, costing $1,695 and $3,195 with 20 and 40
cells, respectively.  These displays will have full cursor
navigation and will work with other computers.  Mr. Saddeek assured
the audience that he, President Chong and Steve Jacobson would work
"to enhance the unit...and make it an exceptional machine."  He
assured the group that he would implement NFB's suggestions.
 
Next, Curtis Chong, a Designer/Consultant for IDS Financial
Services, and Steve Jacobson, a Senior Design Analyst with the 3M
Company, presented a panel discussion entitled "FROM DOS to the
Graphical User Interface, Blind Users in Transition."  In his
introductory remarks, Curtis Chong said that the first place people
will see conversion from DOS to the Graphical User Interface (GUI)
will be in the office. He said that his employer had decided to
make the switch from DOS and the Macintosh to OS/2.  He said that
he thought a useful program item would be to discuss the day-to-day
experiences of switching over or "how we got out of switching
over."

Steve Jacobson, who serves as NFBCS vice president, stated that he
has been in the "computer business" for many years.  He said that
Curtis chong had converted to the GUI while he was barely beginning
the process.  The first question that a blind person needs to
answer is whether conversion to the GUI is immediately necessary. 
Four years ago, the 3M Company decided to convert to the Windows
environment.  Steve Jacobson had to search for DOS options since
adaptive technology to make the Windows environment accessible was
not available at that time.  He discovered that converting to a
Windows environment did not necessarily mean using Windows
software.  Word processing was largely text based, and the terminal
emulation done under Windows could also be done under DOS although
perhaps not as efficiently.  Mr. Jacobson said that he had no
trouble getting files converted to or from Word for Windows
although a secretary might not have that option because formatting
information could be lost due to version skew.  In the case of
terminal emulation (using a PC as a terminal connected to a
mainframe computer), depending upon the type of connection, there
were DOS solutions.  It is definitely worth investigating whether
conversion is really necessary.  Mr. Jacobson said that he is now
investigating conversion to the GUI for several reasons. First,
during the last year, the ability to access OS/2, Windows and their
associated application programs has improved markedly.  There are
more options for GUI access today, and things are going to get
better.  Second, memory management is much easier under OS/2 or
Windows for multiple applications (speech software, terminal
emulators, LAN software, etc.).  Third, although it is not a major
reason, most of us will soon be using a GUI of some sort in the
workplace; it's time we got used to it.  The days of using graphics
under DOS are receding, he averred.  Windows and OS/2 give more
automatic graphics access than under DOS.  If we consumers do our
jobs correctly, we may have better application access than we've
ever had under DOS.  In the meantime, it is probably possible to
postpone conversion to a GUI and still use DOS applications to buy
time until the access technology has matured.

Curtis Chong stated that he works for a unit called Operating
Software and Performance Management.  He coordinates the work of
the Communication Software Support Team.  His main focus is
communications software and data transmission methodologies.  Last
August, IDS decided to switch to OS/2.  When he went for training,
however, it became clear that the local IBM gurus did not know how
to do anything without using the mouse.  The OS/2 manuals were no
better; they did not explain the underlying structure or what files
were being used/changed; everything was in terms of that pesky
rodent.  When he got his OS/2 system, he had to get help from the
Workstation gurus who "treated me like a user."  Finally, Mr. Chong
got a speech program to run under DOS and Screen Reader/2 to run
under OS/2; his PC was set up with the "Dual Boot" configuration,
allowing him to run either OS/2 or plain DOS, depending on his
requirements.  Plain vanila DOS was sometimes necessary, he said,
because OS/2 would occasionally lock certain files needed by some
applications; his print spooler wouldn't work, and he could not
print from the screen.  Christmas vacation was used plowing through
a Screen Reader/2 tutorial, but confusion still reigned.  Finally,
Mr. Chong was able to secure a week's training from Frank DiPalermo
of Ability Consulting; he still gets confused but can "fool around
and do things."  Mr. Chong said his coworkers are jealous; he can
do things under OS/2 with the keyboard faster than they can with
the mouse.

In the question-and-answer period, David Andrews, Director of the
International Braille and Technology Center stated that the
National Federation of the Blind would be evaluating all the
systems designed to make Windows and the GUI accessible to the
blind.  Other persons related their experiences switching to OS/2
or to Windows.  In response to a question about MicroSoft's
Chicago, Greg Lowney of Microsoft stated that documentation on
Chicago had gone to various screen reading program vendors and that
apparently some were finding that it was easier to work with than
expected.  Adequate training for blind computer users (or the lack
of it) was raised as an issue by several people.

The next item was a panel discussion entitled "The Blind Too Can
Travel Along the Information Superhighway"; the presenters were
Lloyd Rasmussen, Senior Staff Engineer, National Library Service
for the Blind and Physically Handicapped; Mike Freeman, Computer
Systems Programmer, Bonneville Power Administration; and David
Andrews, Director, International Braille and Technology Center. 
Lloyd Rasmussen briefly discussed the commercial online service
GENIE; Mike Freeman discussed the Internet; and David Andrews
discussed other online services.

Lloyd Rasmussen said he started telecommunicating some 15 years ago
with a DigiCasette and a 300-baud modem; this taught him much about
serial communications.  He then began talking to a deaf-blind
person, Georgia Griffiths, using a VersaBraille and modem, first
long-distance and then via CompuServe.  Mr. Rasmussen has been on
GENIE since 1987 (it began in 1986).  He likes GENIE because it is
easy to use and there is a wide variety of information and services
available.  For example, one can find news/sports from Reuters' and
many bulletin boards and roundtable discussion areas divided into
categories and topics.  The bulletin boards also have file areas
(from which one can download) and areas for announcements.  There
is a fine disabilities bulletin board, and the Braille Monitor for
the last 2 years can be found there.  One can get the "Cyberspace
Report" in ASCII, the Newsbits service, and there is E-mail access
to the Internet.  Mr. Rasmussen opined that the Internet is just
the beginning of the information superhighway; he recommended
reading books such as "The Big Dummy's Guide to the Internet" and
"Zen and the Art of the Internet."

Mike Freeman then briefly discussed the Internet.  The Network
Information Center (NIC) assigns addresses and names (domains) to
the various networks which comprise the Internet.  There is, he
said, no over-all control of the Internet.  He described the
Internet as "controlled anarchy."  He next discussed the vast
variety of information available via the Internet (Usenet
discussion groups, mailing lists, an untold number of files about
anything imaginable, etc.).  One can gain access to the Internet
through various commercial Internet providers, the online services
Delphi and America Online and through various bulletin board
systems.

Dave Andrews then gave an overview of online services.  He
described the Internet as "geeks gone mad".  Until fairly recently,
he said, the Internet was relatively unknown.  Until recently, at
least, many machines on the Internet used the UNIX operating
system--great for the initiated but confusing for the novice. 
Although many people think of the Internet as the information
superhighway, he maintained that the latter term includes more than
the Internet.  He then described other online services.  CompuServe
has from one to two million subscribers.  He called it the
"Cadillac of online services".  Starter kits are available to make
subscriptions easy.  CompuServe tends to be more expensive to use
than other online services, but "basic services" are available for
a reasonable fee.  There are some 500 "forums" or discussion areas,
many files to download, travel and shopping services and much more
available through CompuServe.  There is software support and
CompuServe is linked into the Internet.

Another online service, Prodigy, is quite slow.  There is to date
only one program in the DOS environment to make it accessible via
speech.  The latest release of Prodigy software may have broken
this.  Prodigy is a pain to use, Mr. Andrews said.  There is big
money being spent here; Prodigy spends on the order of $400 per new
user.  Mr. Andrews briefly mentioned other online services such as
GENIE, AMERICA Online (AOL) and Delphi.  Delphi is currently the
only online service with full Internet access (Telnet, Ftp, etc). 
He concluded by saying that many files dealing with the Internet
are available on NFB'S bulletin board system, NFB-NET.

The next speaker was Dr. James Thatcher, Manager, Interaction
Technology, Mathematical Sciences Department, Thomas J. Watson
Research Center, IBM Corporation.  His presentation was entitled
"Access to the GUI, Hope and Frustration".  Before making his
remarks, however, Dr. Thatcher was given an award by the NFB in
Computer Science for his outstanding work in making IBM PC's (and
most especially, the OS/2 operating system) accessible to the
blind.  Dr. Thatcher thanked President Chong for the award, saying
that he was originally a mathematician; he became involved in
screen-access for the blind through helping a blind colleague with
his computer access problems.  He has been involved with this
subject now for ten years.  Dr. Thatcher said that it has totally
changed his life.  Mathematics is boring when compared to it, he
said.  He again thanked the NFBCS, saying that he'd been coming to
NFB conventions since 1989.

In his talk, Dr. Thatcher said that we should not judge the systems
being developed to give the blind access to the GUI on the basis of
mature technology.  The task is far more complex, and the
environment is many orders of magnitude more so.  What makes things
frustrating is that new operating systems using the GUI are
continually being developed.  However, Thatcher maintained that the
developers of these new operating systems were becoming
increasingly aware of the special access requirements for the
blind.  He mentioned that IBM and the X Consortium were working on
the problem.  He said that he was fortunate that heavy duty
development of Screen Reader/2 had slacked off as he could now fix
problems and add fun things like multimedia support to PAL (Profile
Access Language).  This year, Dr. Thatcher said he started to
develop a toolkit for Screen Reader and to use PAL to make Screen 
Reader act like WinVision and Jaws for Windows.  One of the
problems he confronted was that the various screen reading programs
have very different philosophies as to how things should look and
feel; this makes it extremely difficult to diagnose and solve
problems.  Features and assumptions considered sacred by one vendor
are not considered important by others.  Dr. Thatcher said that
this plethora of program styles must make it very difficult for
evaluators such as NFB to compare the various programs. He said
that articles comparing screen reading systems are often weak in
that they do not describe explicitly what they expect the programs
being compared to do.  In addition, developers address different
questions and concerns when writing about their products.  The
screen readers are so different that we need some common concepts,
methods and tools to understand what they do and how they do it. 
This was not so much the case when we were dealing with text-based
screen-readers. Thatcher proposed a way to get various
manufacturers of GUI screen-readers to talk with one another to
address the issue of trying to understand what the various screen
readers do. He proposed that NFB should conduct a workshop of
screen reader developers, users of Windows-based programs, user
interface experts etc.  The goal of such a workshop would be to
produce an evaluation sheet which would be sent to all developers
of GUI-BASED screen readers.  Such an evaluation sheet would
describe contexts, the tasks users need to perform in those
contexts and what should and should not be spoken (there will be
various opinions on this so screen readers should be able to
accommodate various preferences).  Something on the order of twenty
contexts including some standard Windows applications should be
included.  Screen reader developers would be requested to fill out
these evaluation sheets, outlining the specific steps one would
take using their program to accomplish the indicated actions. 
There would thus be standard material upon which to base screen
reader evaluations. The evaluation sheet could include system
installation and other necessary tasks as well.  Such sheets might
help developers to develop better screen readers and would be a
non-threatening way to get the various developers talking with one
another.

In the discussion following Dr. Thatcher's presentation, David
Andrews said that Dr. Thatcher was a great help in setting up OS/2
at the International Braille and Technology Center.  Mr. Andrews
said that Dr. Thatcher pointed out to him that, whereas with
text-based screen readers, a blind person could examine the screen
fully by going into review mode and examining the screen character
by character, in the GUI environment there were objects on the
screen which sighted persons see which screen readers do not know
about nor represent to the blind person.  The blind do not always
get the whole picture as the sighted see it.  Furthermore, the
blind have no way to know if this is the case. Dr. Thatcher replied
that screen reader developers for the GUI environment have the
problem of missing some things; there are often "unshareable"
tricks performed by applications that tend to confound screen
readers. In addition, each screen reader organizes things
completely differently.  Each developer decides what he/she thinks
the user needs to know.  Mr. Andrews said that most persons want to
get a particular application, run it, manipulate the data and get
out.  On the other hand, programmers may need to know the position
and relationships between icons on the screen.  There are disparate
needs within the blind community.  Mr. Chong said that part of the
problem was that the blind had not had sufficient experience with
the GUI environment to know how material should be presented.  We,
the blind, must garner sufficient knowledge to articculate clearly
how information should be presented in this area.

The next item was the NFBCS business meeting.  Minutes of the 1993
meeting were approved as read.  A detailed treasurer's report was
postponed as the treasurer had left the report in her hotel room. 
She said, however, that the balance was slightly greater than
$1,100.  There were a couple of large expenses during the year. 
President Chong said one was to fly him to Kansas City to help a
blind woman keep her job (which she still has).  There were some
expenses associated with the NFBCS meeting.  There were newsletter
expenses:  $160 for casette duplication and $60 printing costs. 
There was also the expense of the award to Dr. Thatcher.  The
Treasurer's report was accepted as given. After some discussion and
a short recess, elections of officers were held.  The following
were elected for two-year terms:  Curtis Chong, President; Steve
Jacobson, Vice President; Mike Freeman, Secretary; Suzie Stanzel,
Treasurer; D. Curtis Willoughby, Board Member; Jim Willows, Board
Member; and Lloyd Rasmussen, Board Member.

Next came a panel discussion entitled "Designing a Screen Reading
System for Windows."  The presenters were Daehee Lee from Blazie
Engineering and Dale McDaniel, Vice President for Marketing, Artic
Technologies International.

Mr. Lee said that the programmers of Windows Master had three goals
in mind:

1.   The GUI environment should be converted to speech in
     real-time;

2.   Information should be reviewed and spoken only when necessary;
     and

3.   Windows Master should be easy to use.

In accomplishing the first goal, Mr. Lee said that screen reader
designers had to decide what information was useful in the GUI
environment.  Blazie Enginering has decided that four categories of
information are useful:

A.   Text and attributes at a particular location on the screen
     (font, color, etc);


B.   Bit map and icon information;

C.   Mouse information (so that it can be emulated and followed);
     and

D.   The status of the various windows.

In writing Windows Master, Mr. Lee said he applied the
"procrastination principle," that is, analysis of information is
delayed as long as possible consistent with a real time environment
in order to convey the information most intelligently.  In order to
review the screen, the internal representation keeps only
unrecoverable information.  Normally, Windows Master speaks the
information in the activated window.  Screen reading is automatic
and can be configured to handle situations such as that in
WordPerfect where the status line is important but should not
always be spoken.  In review mode, however, Windows Master speaks
everything, allowing the blind person to use Windows in the same
manner as the sighted, Mr. Lee said. Since the blind have
difficulty using the mouse, it is disabled.

Dale McDaniel said that in designing a screen reader for the GUI,
problems are faced that were more easily solved in the DOS
text-based environment.  For example, in the GUI environment, even
when doing word processing, the concept of "columns" may not have
meaning.  There is an area of focus which is often problematic to
determine under some applications. If one can determine the area of
focus, one has access.  There are a limited number of things of
interest on the screen at any given time.  There is some
standardization of application interfaces.  However, Mr. McDaniel
reported that the major problems encountered by screen reader
developers are not so much Windows problems as application
problems.  Applications don't always follow the approved standards. 
They develop their own menu systems or dialog functions and throw
up bit-mapped graphics representations of dialog boxes, buttons,
menu bars, and the like.  Moreover, they will often use the mouse
for access without equivalent keyboard interfaces.  Mr. McDaniel
hopes that some standards will truly emerge soon, for the Windows
environment is here to stay.  He said that most systems purchased
within a couple of years would be using it.  Finally, he said that
access to Windows would improve although there might be some
aspects that would remain inaccessible.

The next presentation was entitled "News from Microsoft" and was
given by Greg Lowney, Senior Program Manager, Accessibility and
Disabilities Group, Microsoft Corporation.  He said that although
Microsoft was initially slow to respond to the problem of access to
Windows for the blind, it had listened and was working with vendors
of screen readers. The problems are monumental, he said, but can be
solved.  Mr. Lowney demonstrated a talking FAX machine which is
part of a GUI-BASED suite of office machines which can be
interconnected.

The final speech was entitled "Seamless Access to Current and
Future Technologies" and was presented by Gregg Vanderheiden, Trace
Research and Development Center.  He said that he was greatly
concerned about information access for the disabled.  When the GUI
was on the horizon, a few wondered how the disabled would gain
access, but everyone let the problem get ahead of them.  There is
the potential of this happening again.  There is a growing amount
of mass produced information technology (ATM's, information
terminals, infotelephones and the like) which are becoming
increasingly inaccessible.  Dr. Vanderheiden said that many devices
to which the disabled formerly had access are becoming
inaccessible.  More and more of these appliances will be in the
home to give access to the information superhighway, he said. 
There is a proliferation of devices using touch screens (they are
flexible) and ATM's for different banks act differently.  Such
systems are easy to use for those who can see, but for the blind
and those with some learning disabilities, they pose problems.  Two
of these problems are:

1.   determining what is on the screens of these devices and

2.   making the choices necessary to obtain information from such
     devices.

Dr. Vanderheiden then played a promotional videotape describing an
information kiosk to be placed in public areas which exemplified
the problem.  He said that the Trace Center has tried to get a
consensus upon how such systems should be accessed but with little
success to date.  He said that until standards are agreed upon,
almost anything could be called "accessible," including systems
that are supposedly designed for the blind but omit information
available to the sighted.  The Trace Center has been investigating
ways of setting up "seamless access"--methods of access that work
for everyone.  Dr. Vanderheiden said that funding is needed for
large scale building of prototypes and gathering of information to
determine the best ways to obtain seamless access.  The need is
urgent.  He cited several examples: CD-ROMS in special and regular
education; an airline reservation system made by Southwestern Bell
in which one uses a picture of an airplane to pick one's seat and
systems in which one books hotels by using maps.  Accessibility is
more readily achievable initially than via retrofitting.

In the discussion following this presentation, the opinion was
voiced that any standards resulting from Trace's work should not
discourage creativity; that the vendors should have the freedom to
try different solutions.  It was also strongly suggested that the
Trace Center should work closely with NFB in the development of any
seamless information access standards.

The meeting was adjourned at 5:27 p.m.

Respectfully submitted,

Mike Freeman
Secretary
National Federation of the Blind in Computer Science


                 ADOBE PDF - ACCESSIBLE OR NOT?
                        by Steve Jacobson

We used to think that getting an ASCII copy of a document was the
final answer.  However, during the past year, things have begun to
get a bit more complicated.  Such developments as the World Wide
Web (WWW) on the Internet and Adobe Systems' Portable Document
Format (PDF) have been drawing a great deal of attention.  Although
not all of the problems associated with accessing the World Wide
Web have been solved, much progress has been made, and many more of
us take advantage of it than we did a year ago.  Adobe's PDF is
still a problem, though.  What makes it a problem, and why should
we care?

Until fairly recently, blind people and computer programmers were
the most avid users of documentation in ASCII form.  Gradually,
though, as more and more people began to use computers, products
became available that tried to make electronic texts function more
like books.  Such products as IBM's BookManager, Wordcruncher, and
Folio's Infobase system, to name just a few, all permit the user to
highlight an item in a table of contents and jump directly to the
text of that item.  Such products often permit you to extract
portions of text and incorporate them into other documents.  But
unlike word processors, they do not permit alteration of the
original document.  These products are often categorized as
"viewers."

Even with their many bells and whistles, the MS-DOS versions of the
products mentioned so far display information using a standard text
screen with 25 lines by 80 characters.  Although the information
can be referenced like a book, and in some ways even more
conveniently, the information still looks like text on a computer
screen, albeit with lots of different colors and pull-down windows.

Particularly in the past year, viewers, also called browsers, have
been taken a step further.  Some of today's viewers and browsers
attempt to recreate some of the appearance of an original
electronic document, including its page format and the size and
appearance of characters on the page.  Headlines and titles may
appear in large letters while book titles might well be italicized. 
It is this technology that the federal government wants to employ
for the distribution of documents and forms--for some good reasons.

In September of 1994, a panel was appointed by the National
Institute of Standards and Technology to examine the various
products and formats that are available to electronically transmit
documents while preserving their appearance and to recommend a
"Final Form Portable Document Delivery Format" (PDDF) to be used by
all federal government agencies.  In addition, the panel was to
include in its criteria the ability to encrypt documents and to
prevent alteration of documents when required.  The format of the
files conveying electronic documents was to be such that viewers
could be written to run on many different platforms.  Finally, if
possible, a commercial product that is already on the market was to
be found to meet these requirements.  Such a document format would
then be defined in a Federal Information Processing Standard (FIPS)
and would become the means of distributing electronic documents.

The product and file format recommended by the panel to NIST is
Adobe Acrobat and its associated "Portable Document Format" (PDF). 
This product meets the criteria set forth by the panel, but it
presents some problems to blind people.

For starters, there is currently no viewer or browser that runs
under MS-DOS.  Apparently, the Windows viewers do work with Windows
access software, but complex documents are still difficult or even
impossible to access.  One user reported that tax forms could be
utilized with a little work, but the online version of the New York
Times could not.

Since Adobe Acrobat is geared to display a page image with titles,
subtitles, and other items identifiable by their associated
typeface or font, the process does not use specific codes to
identify the type of information being displayed.  The viewer
simply displays text in whatever manner the author specifies.  This
lack of structural information can create problems if, for example,
one wants to transcribe the document into braille.  Consider the
problem of separating a page heading that is repeated on each page
from the text.  Consider what happens if text columns vary in width
and the distance between them is changed to allow a picture to be
inserted.  Codes or "tags" could be used to identify such elements
as headings, tables, paragraphs, numbered lists and the like.  Even
though Adobe Acrobat can produce documents that can contain these
structural identifiers, the authors of documents may choose to
denote various types of information solely by varying the font
used.  This seems to be where the most difficult problem lies for
us.

There are other systems for electronically distributing documents
that tend to require better identification of a document's
structure.  SGML (Standard Generalized Markup Language) is one such
system which many feel would provide almost perfect automatic
formatting of documents for braille.  The International Committee
for Accessible Document Design (ICADD) has also defined a set of
codes or "tags" that could, if used, greatly reduce the difficulty
in brailling or enlarging documents.  However, the panel felt that
neither SGML nor the ICADD "tag set" met some of the other
criteria.

Having said all of this, the news is not all bad.  The panel's
report contains some language requiring some degree of
accessibility.  Here is part of that language:

12.  Accessibility of Text, for Copying or Extracting or for
     Indexing (with some level of formatting information) -  The
     textual information contained within a final form document
     should be accessible both to extract the text (with as  much
     information about its format as possible) and to index text
     using a full text indexing technology.  The format should
     allow for conversion to braille encoding or voice synthesis
     appropriate to meet requirements under the accommodation law
     for accessibility to persons with disabilities.  The format
     should allow for textual description of graphical elements
     (e.g., a 2x3 photograph in the top right corner of the page of
     a French Poodle standing  up).

13.  Capture Structure -  In addition to the presentation, the PDDF
     should be able to represent structural information, such as
     paragraph markers, article  threading, the flow of connected
     sections of text, subject indexes, and information in a table
     of contents. The PDDF should also support imbedding of data
     such as EDI, SGML or HTML, access to persons with disabilities
     by using the ICADD DTD (ISO 12083).

In the above extract from the panel's recommendations to NIST, one
sees the words "permit" and "provide for" rather than "require." 
Yet, it would do little good to "require" structure, for example,
when it may not be in the original document.

There are some signs that all is not doom and gloom.  The federal
government is funding the development of an MS-DOS based PDF
browser.  Also, there seems to be a real desire on the part of
Adobe Systems to make their product more accessible, to the point
of committing some resources.  When all is said and done, however,
we will probably still need some third party software that can
intelligently reintroduce some document structure based upon font
information.  Here is one simplistic example.  If the first line of
each page is in a different font than the remainder of the page,
assume that it is a page heading.

One might ask why we don't all simply oppose the adoption of the
Adobe Portable Document Format as a standard.  The reason is that
we would simply not prevail.  This type of information transmission
and viewing is already well established within the commercial
sector.  IBM and Apple have announced that they will be including
PDF browsers on their new computers.  Netscape, a popular software
package used to navigate the World Wide Web, will be including the
ability to browse PDF documents in its products.  A number of large
companies have announced that they will produce electronic catalogs
and brochures in this format.

An interesting development is the announcement by Adobe Systems of
a package that will permit paper documents to be scanned and then
translated to the PDF format.  This is not a bit-mapped image or
"picture" of a page.  Rather, optical character recognition (OCR)
techniques are used, and fonts are assigned so that the original
appearance is preserved.  Adobe is marketing this package to
facilities doing large scale electronic archiving of paper
documents.  It isn't plain ASCII, but it is better than bit-mapped
page images.

The point of touching upon some of these commercial applications is
to say that even if the government did not use the PDF format or a
variation of it, we would miss out on a great deal if we could not
access information encoded with PDF.

The intent of this article is not to lobby for PDF or to minimize
its impact upon us as blind people.  Rather, it is my position that
our interests will be better served in the long run if we use the
leverage of a pending government standard to get as much
accessibility into it as we can.  Let's keep a close watch on this
one.


              ACCESSING THE WORLDWIDE WEB WITH LYNX
                       by Darrell Shandrow

     Editor's note:  Darrell Shandrow wrote this article in
     the spring of 1994 while attending Arizona State
     University.  His major was computer information systems.
     Currently, he is taking some time at the Colorado Center
     for the Blind to improve his alternative blindness skills
     and increase his self confidence. In the fall of 1995 he
     intends to continue his college education in Denver
     studying computer information systems and management.

Introduction

I have been on the Internet for over two years now.  In the
beginning, email, TELNET, and FTP were the only facilities in wide
use, and for the blind, they were great things.  Not  only could
one retrieve vast amounts of  information, but there was nothing in
the way of graphics to hamper  access by the blind.  While the
requirement to remember an incredible  amount of information 
necessary to keep track of the various systems was troublesome, it
worked well nevertheless.

Then the wonderful world of Gopher came along.  Many information 
resources became available on a great little critter that allowed 
you to search for a lot of information stored on hundreds of
computers around the world.  Yes, it was headed for graphics, but
it was  still friendly to blind users of the Internet.  Gopher
provided a choice  between using arrow keys and pressing numbers
that were to the left of  each option.  For many of us, both
selection methods worked equally  well.  For others, due to
problems with terminal emulation, using the  numbers as opposed to
the arrow keys was required.  Using the little Gopher critter was
relatively simple.

Granted Gopher is still frequently utilized to provide information
on the Internet.  However, I began to notice a  disturbing trend. 
Applications were becoming more graphical  in nature.  There have
sprung up numerous internet tools for Microsoft  Windows, Macintosh
computers, and Unix systems running X Windows applications.  I
began to think that blind Internet users were going to have a
problem.  You see, there was this new thing called the World Wide
Web that was beginning to achieve popularity and acceptance.

When I first heard about the World Wide Web (WWW) I began to worry. 
People who discussed it on the Internet and at my college talked
about it in terms  of X Windows and the X terminals down in our
Computing Commons.  I  thought, "Wow, here is another system we are
locked out of until adaptive  technology companies and researchers
get their acts together."  I also  noticed new services coming
online with Web access but no gopher access.  These included the
mtv.com system operated by Adam  Curry and the Palo Alto Weekly
newspaper experiment.  I was worried because I had become dependent
on Gopher and did not know anything about accessing the Web.  Well,
I used  my ability to get around and set about finding a solution
to this problem that would be helpful to blind people. I have been
a member of the National Federation of the Blind for a 
considerable time now--over 6 years (Well, it's a considerable time 
for me anyhow.).  As I wrote a message to adam@mtv.com complaining
about  the problem for the blind using WWW, I remembered that we
have always said in the Federation that we do not allow our
blindness to get in the way of accomplishing  our goals.  We say
that we can compete on equal terms with the sighted.  Keeping this
firmly in mind, I began to think.  I remembered hearing about a
program called Lynx which was supposed to be a text-based WWW 
interface.  I first discounted this program because I figured,
"Well, the web uses hypertext  and this depends on highlighted text
so how can it possibly work for the  blind?"

Well, I decided that it was high time to give it a fair chance
instead of complaining about my inability to access the Web.  I
simply tried running the Lynx program on my UNIX account.  To my
surprise, it worked great!

As it turns out, the Web is fairly accessible to us  after all. 
No, we can't access any of the image files in a meaningful way, but
we can most definitely see the text.  As I mentioned earlier, the
Web  uses hypertext.  It also uses the concept of a universal
resource locator  or URL, which is used to keep track of documents
stored on computers throughout the world.  The URL is a single path
to any given internet resource.  This can include a system you log
into using TELNET, a file you can download using ftp, a document
you can view using Gopher, articles in a UseNet newsgroup or a
hypertext document.  It works like this:

You activate the Web program (Lynx, in my case) and are placed on
the home page of a  default server.  This page not only has
relevant information for that  institution but also points the way
to other resources.  You select items on the Web by placing the
cursor on various parts of the screen  that contain highlighting.
These are known as hypertext links.  When a  link is selected the
relevant document or page is brought up.  You also have the choice
of specifying a specific URL on another machine  directly.

Configuring Lynx

(Note: all text enclosed in quotation marks represents what you are
supposed to type.  Do not type the actual quotation marks.)

In order to use Lynx, you will need to have an account on a
computer that is running the UNIX operating system.  (A version of
LYNX that can run under DOS is also available, but I do not know
anything about this program; my experience is with Unix-based
Lynx.)  Hopefully, the Unix system will already have the Lynx
program installed and ready to go.  If not, you can download and
compile it yourself since it is a free program.  If you are
uncertain as to how to do this, you should contact someone who has
compiled such programs.  You may also contact me for help; my
contact information can be found at the end of this article.

Along with a modem, you will also need communication software that
is capable of emulating a vt100 or vt102 terminal. Most standard
communications software such as Telix, Procomm Plus, and Terminate
can easily be configured to do this.

To run the Lynx program, simply type "lynx -show_cursor" at your
system prompt and press enter. The -show_cursor is very important
as it causes the cursor to appear at the beginning of each
highlighted hypertext link.  Running Lynx this way greatly reduces
any need to track highlight bars or other information on the screen
not directly adjacent to the cursor.  Consequently, you should not
have to make any special configurations for your screen reading
program.

Using Lynx

As I said earlier, Lynx is text-based.  Nevertheless, it supports
hypertext links and all the concepts that come with them.  It
allows  you to select links by pressing the up and down arrow keys. 
The cursor is placed on the first letter of the current link.  The
up and  down arrow keys are used to move between links.  There are
often several links displayed on a  single line.  The up arrow key
moves to the left and goes to the previous line while the down
arrow key moves  to the right and goes to the next line.  The user
presses either the right  arrow or the enter key to select a link. 
The left arrow key is used to  go back up to the previous logical
level.  This would be where you were before  selecting the link.

Reading links is not difficult.  Remember that the cursor is placed
on  the first letter of the link you are pointing at.  You simply 
direct your screen access software to read the current line.  For
situations where there are several links on a line, you can get
more specific  by reading the current word under the cursor.  This
word is the first word of the  link being pointed at.  Since
hypertext links normally consist of a couple of words each, it  is
easy to figure out which words on the current line constitute your
link.  Once you figure out this process, you will have no trouble
reading lines  that consist of three or more links. A braille
display makes this task much easier.  Since at least a quarter of
the line is on the braille display at one time, you can feel where 
the cursor is located and the words following it. Often, the link
is  right under your fingers or one advance of the display away
from  complete visibility.  Of course, the braille display's cursor
tracking function  must be active when using Lynx in order to have
this benefit.  I  recommend, when humanly possible, the use of both
speech and braille when  using Lynx.

To "Go" to another URL press the "G" key.  There will be a prompt
asking  where to go.  You must type a URL in the form:

http://uu-gna.mit.edu:8001

and press enter.  This is a realistic example.  If you have Lynx
(or any  other web browser for that matter), try this.  The URL
points to an  MIT computer, and one of the links is the Globewide
Network Academy.  Move  up or down to find this link and press
enter.  Select various links by  moving, reading the current word,
and pressing the enter or right arrow  key. Remember that the left
arrow key always sends you back up one level.  Don't worry.  It is
difficult to  mess up WWW.  When you are ready to quit, simply
press the "Q" key.

Feel free to email or call me at  anytime.  Here is my contact
information:

Darrell Shandrow
c/o Colorado Center for the Blind
1820 S. Acoma
Denver, CO 80223
Phone: (303) 778-1130 (days)
       (303) 758-2904 (evenings)
Internet: nu7i@netcom.com