Books for Blind Scientists: The Technological Requirements of Accessibility
This paper describes three new developments that hold great promise for improving the accessibility of scientific literature for people who are visually impaired or who have significant vision-related learning disabilities. All rely on the availability of information in high-level electronic form. A brief review of methods for storing high-level information and an example of their use in printing Dotsplus documents are given.
Computer technology is revolutionizing information accessibility, and few people have benefitted more than those with severe visual impairments. Voice synthesizers and computer Braille displays allow blind people to read computer text as easily as sighted people do. Computer Braille translators allow Braille documents of a large amount standard literature to be printed directly from word processor files. Computerized reader services provide news, weather, sports, stock reports, and other information by telephone.
Unfortunately easy access to computerized information has been largely restricted to words. The information revolution has barely touched the problem of access by blind people to math and science, fields whose literature cannot be represented adequately by words alone.
Blind people presently have only three realistic options for reading literature containing scientific notation - human readers, tape recordings, or Braille. Because human readers are expensive and seldom reliable, high quality tape recordings are generally preferable. Recording for the Blind, a nonprofit company based in Princeton, New Jersey, maintains a large library of textbook tape recordings. For many years RFB has been a critically important source of accessible textbooks for many blind students. RFB will record any book on request, but a lead time of many months is required. Tape books are most useful for materials that can be read from beginning to end. For science and math books in which the reader must frequently refer to equations, tables, appendices, etc., tape recordings are a poor substitute for a book.
Standard Braille is inadequate for science and math. It does not even have symbols for "plus" or "equals." A special "math Braille code" is used for math, but it is not really adequate for such subjects as chemistry, physics, or engineering. Unlike printed mathematical and scientific notation, which is more or less internationally standardized, the Braille math codes used in the United States and the United Kingdom are different, and both differ substantially from math codes used in non-English-speaking countries. Most blind people do not read Braille, and only a very small percentage of those people read math Braille. Blind children find it particularly difficult to learn math because they must first learn a separate math Braille language.
Another problem is that almost all literature involving scientific notation must be Brailled by hand, and very few Braillists know the math code. Only the most common K-12 math and science texts are routinely translated. Other texts can often be obtained only by contracting with a Braille transcriber. This is expensive and requires along lead time.
Three new developments can potentially revolutionize accessibility to math and science literature by people with severe visual and learning disabilities. None of these technologies can provide ready access to printed books, but information that is available in high-level computer files can be readily accessed.
One of these new technologies is the AsTeR computer program (Raman 1992, Barry 1994) (described in this journal by its developer, Dr. T. V. Raman. A World Wide Web (WWW) demo of AsTeR is available at:
Dotsplus, a method for producing hard copy scientific literature, is another important development. Developed at Oregon State University, Dotsplus is described briefly later in this article. Visual examples of Dotsplus documents are given in Gardner (1993), and are available to Internet users at a World Wide Web site:
Tactile examples are available from the authors on request.
Finally, the utility of Braille for scientific literature should be greatly enhanced by a fundamental Braille reform being considered by the International Committee on English Braille (ICEB). A committee appointed by ICEB is presently developing a proposal for a single unified Braille code based on an expansion of the present Standard English Braille literary code.
The unified Braille code project chair is Ms. Darleen Bogart LIB-DARLEENB@immedia.ca. The Braille Research Center, which is directed by Ms. Hilda Caton, chair of the Braille Authority of North America, has a list server, BRCTR@ULKYVM.LOUISVILLE.EDU, that reports on unified Braille research developments. The list moderator is Dr. Emerson Foulke, email@example.com.
The authors' Dotsplus research group is developing practical methods for making these three new possibilities into realities. Future technological developments should make it possible for people to have Braille as well as voice output from AsTeR, or simultaneous Braille and audio. Many sighted people may find it useful to have both audio and visual information presented simultaneously. In the longer term, it is possible that high resolution tactile output devices will permit blind readers to "see" graphs, line drawings, etc. Specialized portable computers with full-page, high-resolution tactile display and voice output could finally give blind people both the accessibility and portability of print books. A number of innovative research efforts are underway to develop such tactile displays (Leeb, 1994, and Fricke, 1994).
None of these developments will provide easy accessibility unless the written material is available in a high-level electronic form. It is possible that electronic publishing will someday make all materials available in such a form, but today most authors and publishers regard computer files as little more than a convenient intermediate step in the process of publishing a book on paper.
HIGH LEVEL ELECTRONIC INFORMATION
All printed books are, in essence, picture books because they are visual illustrations of something. A math book, for example, commonly contains printed letters, numbers, other scientific characters, flow diagrams, line drawings, and occasionally pictures with varying shades of gray or color. The reader's eyes and brain translate the pictures of characters into meaningful words, sentences, tabulated information, equations, etc. Line drawings, graphs, flow diagrams, etc. also are visually interpreted as information. Almost everything appearing in a scientific book or article is intended to convey quantitative information to the reader.
A book can be introduced into a computer by scanning it with an optical scanner and storing its pages as bit-mapped images. Information in such a "low-level" file can be accessed visually by displaying it on a computer screen, but this information is no more useful to a blind person than the original book. Artificial intelligence computer programs can often recognize letters and numbers in such files and make that information available in other ways. For example, a number of commercially available reading machines scan pages and use a voice synthesizer to read the pages aloud to blind users. Many can also make computer files that can be printed in Braille or read on other computers using either a voice synthesizer or a Braille display. Unfortunately these reading machines are slow and work well only for the simplest kind of printed material. The computer recognition routines can be confused easily if the print quality is poor, if several font types are used, if the page contains columns, tables, figures, or other complications that are common in much literature. In addition, there are presently no commercially available programs capable of recognizing any but the most common scientific symbols, and all are incapable of reconstructing a scientific equation. It is far better for a blind reader to have a high-level file in which the information, not a picture of the information, is stored in the computer in an easily retrievable format.
One such high-level language is The Standard Generalized Markup Language (SGML)--actually a family of high-level markup languages--which is used by a number of publishers. SGML files contain tags identifying the function of the file elements. These tags are described by document type definitions (DTD's) written in the SGML meta-language. Hyper-Text Mark-up Language (HTML) is a DTD developed for the WWW electronic information network. A DTD developed independently by the International Committee on Accessible Document Design (ICADD) is almost completely included in HTML, and future HTML releases will include the ICADD DTD as a subset. HTML does not presently store mathematical, scientific, or tabulated information at a high level, but new or improved extensions are being developed to include these. When these extensions are completed, HTML files or files using other DTD's that have math tags could provide the blind reader as much information as WWW or a book printed from these files provides to the sighted reader.
At present, the family of mark-up languages based on TeX which are used for publishing many higher level scientific books and articles, can provide excellent high-level files that are in principle easily accessible by AsTeR, Dotsplus, and Braille users.
Unfortunately, the majority of today's publications are still being printed from low-level files such as PostScript, which discards the information inherent in equations, tables, etc. and presents them as pictures. People with severe visual impairments and many people with learning disabilities are effectively excluded.
PRINTING DOCUMENTS IN DOTSPLUS
Dotsplus provides a particularly straightforward example of how high-level information can be reformulated to be accessible to blind people. A Dotsplus document is laid out much like the corresponding print document. Dotsplus symbols are larger by a factor of approximately 2.5, but most scientific symbols (plus, minus, equals, parentheses, brackets, summation sign, integral sign, etc.) are raised images similar in appearance to their print equivalents. Some are emphasized to make tactile recognition easier, but all are instantly recognizable by sighted readers. Letters, numbers, and a few symbols that are difficult to recognize actually are reproduced in an eight-dot Braille code. Symbol position is the same as it would be in an ink-print version. Thus the Dotsplus Braille code must be one in which the cell shape without reference to position is sufficient to identify the symbol. The lower case Dotsplus letters are standard Braille. Upper case letters are those used in most eight-dot codes. Numbers have previously been represented by the European computer code convention but will be represented in future versions by symbols that are more consistent with the unified Braille code.
Dotsplus documents can presently be printed from LaTeX files and from some graphics-based word processor files. Original files maybe viewed and edited on a graphics screen or printed for sighted viewers. Sighted editors can view the document using non-Braille fonts of the appropriate size. Except for the font size, such documents look like any other scientific document. A straightforward global font change converts these into pictures in which the letters, numbers, etc. are Braille dots. This document is then printed by a modified wax-jet printer that produces tactile images, or it can be printed on a special "swell" paper using almost any standard printer. The swell paper is then fed through a machine that heats the paper causing the black portions to swell. In either case, the output is tactile and has proven to be easily readable by good Braille readers after a minimum of instruction and practice. Widespread use of Dotsplus awaits introduction of good quality, commercial wax-jet printers and/or lower priced swell paper.
Dotsplus transforms the visual layout of a document to a tactile layout. This has the advantage that the information conveyed by the spatial layout (eg. a raised symbol is a superscript/power) is the same in a Dotsplus representation as in the visual one. The only high-level requirement for Dotsplus is that fonts must be easily alterable.
Dotsplus translation is not possible from files made by every word processor or mark-up program because not all such programs allow fonts to be changed globally. Programs that retain equations as bit-mapped pictures are inadequate. The equation editor in Word for Windows, for example, does not permit character fonts within equations to be altered without rewriting the equation in the desired (Braille) font. However if the equations are written using"field codes", Word fonts can be changed. The Dotsplus group has made a number of Dotsplus documents from Word files written using field codes.
Braille and AsTeR cannot use spatial representation to convey information and consequently require higher level files than does Dotsplus. Most modern word processors do not presently store information at a high enough level, but future versions are likely to include options for transforming files to TeX and/or SGML. Several shareware and commercial products that allow easy authoring of TeX and SGML are either presently available or have been announced for future release.
Widespread use of software-producing, high-level files and procedures for making such files available to individuals with disabilities are required before math and science can be made truly accessible. Even then, more research on non-visual representation through sonification, virtual reality, and various tactile/audio methods are needed in order to make graphs, figures, line drawings, flow diagrams, and more complicated non-textual information readily understandable by people with visual disabilities. Such techniques will, of course, be useful in practice only if the information, not just pictures of that information, are stored in the computer files.
* Research supported in part by the National Science Foundation.
_Word for Windows_ is a trademark of Microsoft Corporation.
_PostScript_ is a trademark of Adobe Systems Incorporated.
Barry, W.A., Gardner, J.A. and Raman, T.V. (1994) Accessibility
to Scientific Information by the Blind: Dotsplus and AsTeR could
make it easy. To be published in _Proceedings of the 1994 CSUN
Conference on Technology and Persons with Disabilities_, held in
Los Angeles, California.
Gardner, J.A. (1993) Dotsplus - better than Braille? Published
electronically in _Proceedings of the 1993 International
Conference on Technology and Persons with Disabilities_, held in
Los Angeles, California.
Fricke, J. (1994) Presentation at the Symposium on High
Resolution Tactile Graphics, Los Angeles, CA, March 15, 1994.
Leeb, S. (1994) Presentation at the Symposium on High Resolution
Tactile Graphics, Los Angeles, CA, March 15, 1994.
Raman, T.V. (1992) An Audio View of La(TeX) Documents. _Proc.
TeXUsers Group_, 13, 372-379.