Braille Info

For those interested, and after being badgered by Jason White, I have hosted the source of a braille translator.

And after being prompted by Joe Sullivan, I have added a very brief history of computer based braille.

History and Acknowledgements

Please keep the "acknowledgments" file with the released materials, so that it stays with copies made by others, much as would be the case with the "references" part of a paper. I suppose that is normal for software released under the GNU license, but I don't know the precise customary form, if there is one.

RNIB in the UK have given permission for this software to become open source. On posting it, Joe Sullivan kindly provided the following history of it.

The original DOTSYS program was itself released to the public, more or less automatically as it was developed with public funds. [MITRE is a not-for-profit, especially chartered by Congress to work on projects of direct interest to the government, usually for the Air Force -- but DOTSYS was developed as a research project for the Atlanta (Georgia) Public Schools, in collaboration with the Massachusetts Institute of Technology. For years, one could easily obtain a magnetic tape, typical medium of the mag-tape era, with all the sources and so on, directly from MIT -- maybe it's still possible, for all I know (though I would doubt it -- and who could read it?).]

I'll try to set forth the basic facts for you, though surely these need not be recounted in detail

The DOTSYS Project was jointly carried out by MIT and MITRE, with MIT developing the "hardware" (eventually resulting in the "BrailleEmboss" embosser) and MITRE the "software" part of the system. The purpose was to support the Atlanta Schools so that classroom material could be brailled on a quick-turnaround basis for mainstreamed students, Atlanta being at that time (circa 1969) just starting to implement the mainstreaming of blind students.

Robert A. J. Gildea (who is blind) was the project leader at MITRE. His first co-worker was Dr. Jonathan K. Millen, who developed the theoretical basis for the translation process still used (albeit with modifications) in both ITS and our Duxbury Translator -- first published in "Finite-State Syntax-Directed Braille Translation", Technical Report MTR-1829, MITRE Corp., Bedford, Massachusetts, July 2, 1970. He had also decided on COBOL as the implementation language, in a report dated December 1969, and produced a prototype translator called DOTSYS II, described in a report dated July 1970. (Interestingly, all of us who worked on the project disliked COBOL as a language because of its clumsy verbosity; my usual language at the time was PL/I. But for portability in 1969, COBOL was a very logical choice, and portability was deemed to be of paramount importance.)

I joined the project in 1970 and was chiefly responsible for programming thereafter. W. Reid Gerhart also joined around the same time and assisted with the programming. The resulting released program was called DOTSYS III, described in "DOTSYS III: A Portable Program for Grade 2 Braille Translation", Technical Report MTR-2119, MITRE Corp., Bedford, Massachusetts, May 14, 1971.

Later, at the first International Workshop on Computerized Braille Production, held in Muenster, Germany, March 1973, the work was also described in a paper authored by me, entitled "DOTSYS III: A Portable Braille Translator".

I have copies of those old papers, including the full proceedings of the Muenster conference, though unfortunately only in paper form. Not surprisingly, we ourselves acknowledge influences (though not actual code) from much other work of the era, including from the earliest braille translator project -- a proprietary system developed at the American Printing house in the early sixties.

From there, DOTSYS III did achieve what could be considered widespread use, considering that it was still the "mainframe" era of hugely expensive room-filling machines, and braille was of interest to only a small minority. I believe we ended up counting about 30 organizations that took it up and (so far as we heard) successfully installed and ran it on widely varying brands and kinds of computers, as we had planned. In a few cases, the local staff decided they wanted to work further with it, and re-programmed it, usually in some language better liked by the programmers there. IBM did that, producing a PL/I program called "Braillemaster" that they freely gave away. (There's no relationship, so far as I know, with the current "Braillemaster" product from Robotron.)

RNIB was one of the organizations that adopted DOTSYS and reprogrammed it for their own needs. I believe that John Gill, who is still there, was involved in that effort -- and the language may have been changed to PL/I, though I am uncertain about that detail. No doubt there were many improvements, as well as corrections necessary because of the differences between American and British usage, introduced at both the RNIB DOTSYS stage and the later ITS stage, when the language was changed to C. Here is where you may have some fun discovering the significant history within RNIB itself.

When we founded Duxbury in 1975, we too reprogrammed DOTSYS (in FORTRAN -- a fairly unsuitable language for this kind of task, but that's what we had on our Data General minicomputer). We later produced versions in Z80 assembler code and ultimately C, as we use today. Through all that, the basic algorithms and tables, despite some improvements in both form and function, have remained remarkably stable. So it is not surprising to see that the same has been true in the RNIB branch of the "family tree". (For example, in DOTSYS, the arbitrary number 6 is associated with the "input class" associated with contractions used at the start of a word -- it's still the same number in both ITS and DBT!)

Well, I see that I've got carried away with historical details that surely no one but old fogeys like me care about. But thanks for listening, and please feel free to summarize or quote anything you like.

Incidentally, though it is not dated, I would guess that the paper cited by J. White was written sometime in the late 1980s (based upon the fact that the latest reference in the paper is to one of the author's theses, dated 1985).

Also of possible interest: There was a series of conferences, starting with the one in Muenster in 1973, especially on the subject of Computerised Braille Translation -- in Copenhagen (1976?), London (1979), Toulouse (1981), Wintherthur (1985), and Leuven (1990). I didn't go to Copenhagen one but get to the others, and so have all the proceedings -- lots of papers, for what it's worth (probably not much, in current terms). But if anyone really wants to research braille translation algorithms that have been tried, no doubt with varying degrees of success, there is probably someone at RNIB with a similar collection.

Again, thanks for your message, and for humoring an old history buff.

Best regards, Joe Sullivan. <joe@duxsys.com>

Quite a story Joe. Many thanks. DaveP, December 1999.