I created the first Irish language PC word processor in March 1983. The work began one Monday morning and the resulting product was featured on the TV news just seven or eight days later.
I was working for Sord Computers in Santry at the time. The company had grown to be the second largest microcomputer maker in Japan after Fujitsu, but was still a small firm by Japanese standards. In that era the architecture of personal computers differed considerably from one vendor to another. Sord never got the hang of industry standards, especially when it came to operating systems, but its hardware quality was good. The company also promoted the Pips language, which enabled end users to create their own rudimentary databases and spreadsheets.
Sord had established its first subsidiary company, Orange Computer Products, to create a European base in Santry. The name of the new operation was supposed to signify that it would soon challenge Apple. It opened in December 1980 as a repackaging and distribution centre. The company intended to add system assembly at a later date, but that never got going.
Sord’s first two engineers in Ireland joined in 1981 and five more were hired in summer 1982. Donal O’Mahony and I came from Trinity College Dublin, where I had just completed an engineering degree with a final year project on microprocessors. Richard Holbrook, John O’Keefe and Eamon Thornton came from University College Dublin. I had received other job offers that would have taken me to the United States, but I chose Sord because it would provide an opportunity to work in Japan at a time when the electronics industry there looked exciting. I worked for Sord for 2.5 years and spent half of that time in Japan.
Sord engineers in Chiba, Japan (l-r): Donal O’Mahony, John O’Keefe, James Mahon, Richard Holbrook and Eamon Thornton in December 1982.
Photograph by James Mahon
Everyone in the company was young – the older people were about 30 – and made things up as they went along. They worked long hours and some of them slept under their desks. Sord’s engineers in Japan were like native speakers of machine code for the Zilog Z80 processor, but there was never much software for their systems. When they wanted a word processor, for example, they simply got hold of a Wang manual and produced one thorough reverse engineering.
I worked on a number of operating system and software porting projects, including the first C compiler for a Sord machine. One of my early assignments in Japan was to adapt the word processor for the company’s Intel-based models.
The subsidiary in Santry shipped thousands of computers to customers in Europe and many of these users were in France. Sord’s French distributor (GEPSI) had commissioned modifications to Sord’s FDDOS operating system – the FD stood for ‘floppy disk‘ – so that it had the ability to create accented characters as a keyboard input function, and similarly, could map a single character to a string of characters for printer output. These tables eased the task of European language integration considerably.
The manager of the Santry operation, Yoshiyuki Hayashi, spotted another opportunity to make use of this capability. The Computex show in the RDS was scheduled to open on 28 March 1983 and Sord had booked a stand. He told me to make an Irish language version of the English word processor for a demonstration there on the SORD M23. The original software had been written in Z80 Assembler (in Japan) and could run in a 64K memory machine.
At first I assumed that he was joking. But he was serious and gave me a PROM writer to create the new accented characters. The system already supported lower case vowels in French. But we needed two fadas for Irish – one for upper case and one for lower case.
First, then, I built tables to enable character input of all ten accented characters and output of them to a NEC Spinwriter by mapping to <Vowel, backspace, Fada> and so on. We had to pick 8-bit characters for the accented characters, and then draw them on graph paper and then convert them to binary and load them onto the character ROM and burn it and replace it in the system.
Then, we had to convert the text strings for menus in the OS into Irish. I knew of a guy in University College Galway who could provide Irish versions of technical terms. I asked my old primary school teacher for the rest of the words.
The expert in Galway dictated the terms to me over a poor telephone line during a thunderstorm on the Friday afternoon before Computex. I could not hear the spellings clearly, which resulted in a number of mistakes.
We re-assembled the code and were able to demonstrate it on Monday or Tuesday in the RDS. The main menu (in Irish) was shown on the RTE 6 o’clock news that night. No one seemed to notice the mistakes caused by the thunderstorm.
Over the next year or so we sold about half a dozen systems with the Irish word processor to Udarás na Gaeltachta in Spiddal. This capability also helped us to open an account with the Local Government Computer Services Board.
Although the Sord project was not prompted by the same concerns as the software localisation services that appeared some years later, it required a similar mix of engineering and translation work.
I was in Japan when Sord’s engineers saw an IBM PC for the first time. Some time later we saw an Apple Lisa at the ‘Data Show’ exhibition in Tokyo. It was obvious that the microcomputer industry was about to change. Accented characters might have been a big deal at the start of 1983, but not for the next generation of systems.
Sord was sold to Toshiba in 1985. Both companies expected to transform it into the PC division of the larger corporation. But Toshiba engineers had also developed a portable computer (the famous T3100) and it was this, not the Sord technology, that evolved into a major product line. Sord gave the Santry operation to its local management, who agreed to provide continuing customer support.
I had left the company before the takeover and returned to Trinity to write a masters thesis with financial support from Digital Equipment in Clonmel. My research focused on machine vision – a technology that I went on to work with for more than 30 years.
Last edit: January 2018
© James Mahon 2018