There are pieces of the history of reading machines for blind people that were never told. They were not supposed to be secret, and the blame belongs to many, including me. Having used, tested, and taught 14 reading machines for over 30 years, I hope not to carry these "secrets" to my grave.
I believe that reading machines are not all they should be and all they can become. It makes me wonder how many other potentially valuable ideas fall through the cracks of research projects and human circumstances. In this article, I attempt to tell the bare essentials of what is missing in our reading machines and computer access technology— the hidden part of the history of reading machines—and to suggest how we can supply those essentials in the future. Many of the events to which I refer here happened in the late 1970s and early 1980s. Most of the developers and evaluators did not envision that any of the machines of that day would have a permanent place as tools for visually impaired people.
The Scene in 1980
In 1980, three machines were on the scene, but the Optacon was the only one in use by more than a handful of blind people. Its columns of vibrating reeds presented letter shapes as you tracked the print. High hopes and exaggerated claims in the 1970s had left a backlash of people who were disappointed at the long training needed to use the Optacon, the difficulty of learning to use it, and the price. The blind public grew disillusioned and all but forgot how useful the instrument had become to several thousand of them. These people were reading much of their mail, checking their printing, and doing other small-quantity reading tasks that the sighted public takes so much for granted as barely to consider them reading. A few people could read more than 60 words per minute, so they used the instrument for intermediate-quantity tasks like reading articles and studying manuals. The Optacon is no longer being produced, to the deep disappointment of many of its users.
Caption: Optacon in use, 1970
The second reading machine that was used by only a few people in 1980 was the Stereotoner, which rendered the shapes of letters as a column of 10 auditory tones. It was used for the same tasks as the Optacon, and although it was less costly, fewer people attained speeds above 60 words per minute. Although I may be the person who was most influential in recommending the development and use of the Stereotoner, my colleagues at the Hines, Illinois, Veterans Administration rehabilitation center and I, who were testing adaptive computer technology, never recommended it for wide use. We saw it and the Optacon as transitional and as components of a larger reading system. By 1980, the Stereotoner was no longer produced, and if people think of it at all, it is considered a curiosity at best.
The third reading machine that appeared in 1978, after several years of incubation, was the early Kurzweil Reading Machine. Although the first machines were large, they had only a tiny amount (48 kilobytes) of computer memory. They scanned automatically and used a computer to present synthetic speech to users. The cost was high, and the computer was too small to do the job well. A few were placed in libraries, and the Veterans Administration issued some despite the objection of myself and my colleagues, but few of them were used productively.
By 1989, personal computer memory had "grown" exponentially. The first useful automatic reading machines were produced that year by Kurzweil and Arkenstone. Since then, a stream of small improvements has added substantially to their utility and efficiency. It should be noted that the talking reading machines we now have are commercial optical character recognition (OCR) machines for office use, with added features to make them "friendly" to blind people. This hybrid nature makes them useful and more affordable, but they are also why the machines are a source of disappointment and frustration. They do not have some of the major features that visually impaired people need.
Computers versus Reading Machines
Access to computers gave blind people the ability to read what we typed and much else that we needed to read. It thereby removed at least half the reasons for learning to use the Optacon of that day, sealing its fate as an isolated tool. The trouble is that although people who were skilled at using an Optacon could read most of their mail, long items like books and magazines were impractical to read. People who have an OCR machine (either stand-alone or attached to their computer) can read lengthy items like most books fairly well but not most of their mail effectively. When the computer misses something important, which often happens, one must remove the document from the scanner and either show it to a sighted person or search for the passage with the little Optacon camera. (I usually use a Stereotoner.) That procedure perpetuates dependence and is much more time-consuming than it should be, considering all the investment in equipment, training, and effort.
It is no wonder that people no longer wanted to learn to use the Optacon. It also explains why people who can use it well do not want to give it up. The combination of the Optacon and a computer with a scanner is helpful, but it is like having a happily married couple live in separate quarters. I assure you that most of us who worked hard on the development and testing of reading machines in their early days did not plan things this way.
How We Planned It
Some people accused my colleagues and me of being sadistic and masochistic. Some just hated us for telling what we believed is the truth. The truth is that we never expected people to read exclusively with the machines of 1980. But we questioned then, and we question now, whether computers of the foreseeable future can handle highly degraded print. We were quite certain that computers would not act exactly like a sighted reader and describe bills, forms, and the fancy creations of artistic printers and desktop publishers. After all, print is designed to be read with the eyes. We wanted the Optacon and Stereotoner for those tasks; that is, we wanted a means of examining shapes ourselves, but only when necessary.
Nor did we expect to scan entirely by hand as we must now do with these small machines like the Optacon and the Stereotoner. (Print is not made to be tracked by hand.) We envisioned the small machines integrated into our computerized ones. We wanted to be able to have the image of the page scanned into our computer, which is now the case with the large stand-alone reading machines and the computers. We could then choose automatic reading or take a "closer look" by switching to manual tracking, which would allow us to examine a passage or the layout of a page more easily by using a mouse or puck on a tablet. Then we would have choices. We could hear spoken words, hear tonal shapes, feel vibrating patterns, feel a matrix of raised dots, see a screen with large print—whichever ones the individual needed. We wanted to choose either guided or freehand tracking.
We expected a greater range of zoom capacity than we now have with the small machines. Larger shapes and page formats would be more accessible and not as difficult to examine as they are now with current small machines. We also wanted access to computer screenfuls of information, including the ability to examine the shapes of icons in the same ways.
We also wanted software, the beginning of which is being implemented in the latest versions of reading machines. We planned that once the image of a whole page or screen image could be stored in computer memory, which is now the case, software would be written to describe the layout or format of a page in much the way a sighted reader can. We could then select the titles, columns, or captions we wished to read. As noted, we also hope in the future to choose the media in which to examine them.
Several attempts were made to implement the plans I just listed, but they failed because of the small memories and slow processors of their host computers at the time. There was the Cognodictor, the Talking Optacon, and Raymond Kurzweil's effort to do it in his second computer, the 64-kilobyte desktop model of 1980. There were others of which you probably never heard. You may be surprised to learn that in 1971, the first computerized reading machine that worked was built. It had eight kilobytes of memory and delivered spelled speech.
Once we realized that the personal computers then being introduced had far too little memory, some of us urged that the job be postponed until the1980s. Then, when personal computers "grew" to hold 50 megabytes or more of memory, as we assumed they would, our reading machines would be "ready to fly" for us. We could not get support.
Learning to use the reading machine I propose would require effort, but it would not demand the high level of skill required to use the Optacon or Stereotoner. The payoff would also be greater. For one thing, building reading speed would be optional because users would usually choose to examine only short, "problem" passages with the tactile and tonal outputs. The effect of adaptation—the reduction of sensitivity in the fingers that occurs after continuous exposure to a vibratory output—would also be reduced. And people would be free to choose only a speech- or braille-display output if they wished. Everyone, however, would gain the advantage of the format descriptions I described. That is an improvement in software that can and should be added to current designs.
This leaves us with some important questions. If the machine we wanted then and which I now propose is developed, would it sell? Would people who are blind and our educational and rehabilitation systems invest in the equipment and the necessary training? The amount of investment will depend on the options chosen and whether development costs are to be recovered by the developer. I believe that a sufficient number of people would choose it and that the market would be worldwide. The hardware would be functional in any language.
Where Do We Go from Here?
Some further investigation and testing remain to be done. If it is successful, the question will be this: Will visually impaired people ask for and work for the reading machine and computer-access tools we really need? I hope so.
The personal computers are now big. The goals described here are no longer long-range ones. They are achievable in the short range. Also, more of the blind public now has a clearer understanding of what we can and cannot do with computers.
For example, I am tired of hearing it said that technology is our friend. We have learned that it is the people who sponsor, develop, test, market, teach, and help each other with technology who can be our friends. We now know that computers do not work with us nor play with us; we work and play with them. They can be our tools and toys. We are told that machines can read to us. We have found that machines do not read to people. It is people who read with machines.
We need a reading tool that is configurable in different ways. A young blind child should be able to use it. Blind adults should be able to read their mail with it even if we have to work at it. Blind engineers should find it challenging and useful. If we expect merely to be played with and read to by a machine, then perhaps developers will give up trying. Or if the right product is developed, then maybe blind people will not bother to undergo the training or even to study the manual. In short, we have learned that there are trade-offs that we will ignore if we expect too little or too much.
Developing this reading machine will not be easy, and it will not be cheap. Do blind people feel worthy enough, ambitious enough, and desperate enough to take action? Will we work and advocate for the development hard enough to get it done? I believe we will if we learn the secret history behind the development of reading machines.