1. The JW-10: Japan's First Word Processor, Desk-Sized and 6.3 Million Yen
The JW-10, announced by Tokyo Shibaura Electric (now Toshiba) in September 1978 and shipped in February 1979, is considered Japan’s first Japanese word processor [1]. It was the size of a standard office desk, weighed 220 kg, and had a price tag of 6.3 million yen [2]. At the time, the starting salary for a university graduate was about 100,000 yen, meaning the machine cost the equivalent of 63 months of salary, or more than five years’ wages [2]. This was not a personal item but a machine purchased by companies for document creation. The question remains: why was such an expensive device necessary?
2. Why Early Computers Could Only Handle Katakana, Not Japanese Writing
Japanese writing mixes hiragana, katakana, and kanji. However, around the time the JW-10 was built, computers could generally only handle katakana characters, except in some special systems [1]. Japanese typewriters existed but did not become widely popular, according to reports [8]. An ITmedia article explains that dedicated word processors were born as devices to create documents quickly and beautifully [8]. The challenge was how to create mixed kanji-hiragana text using machines that could only handle katakana. The solution lay in separating the input process from the display process.
3. How the JW-10 Chose the Right Kanji When Converting Kana
The JW-10 could convert text entered in kana into mixed kanji and kana text [1]. Users entered the reading of a sentence by "bunsetsu," which are natural chunks of a sentence like "watashi wa" (I am) or "hashi wo" (chopsticks). When multiple words shared the same reading (homophones), the system used surrounding context and the frequency of word usage as clues to select the kanji [2]. For example, the word "hashi" could mean bridge, chopsticks, or edge. If the surrounding words included "eat," it could narrow the choice to chopsticks; if they included "river," it could narrow it to bridge. While humans understand this through flow, the machine relied on word connections and usage counts. The dictionary contained 54,000 general words and 8,000 proper nouns, totaling 62,000 entries [2]. Development was not smooth initially; early experiments estimated a conversion rate of only 50–60% for actual office documents because many words were missing from the dictionary [3]. However, improving the dictionary significantly increased accuracy in the final product [2].
4. How the JW-10 Let Users Correct, Insert, and Save Text
The JW-10 allowed users to correct entered text and insert new text in the middle, an advantage typewriters lacked [2]. Storage capacity was also notable. Counting 40 characters by 40 lines as one page, the internal 10 MB hard drive could hold 200 pages, while an 8-inch floppy disk could hold 60 pages [2]. The 200 pages equated to 320,000 characters. The source materials do not explicitly state the reason for the high price, so this is speculation: the ease of rewriting and the ability to save text may have encouraged such a large purchase.
5. Fujitsu's OASYS 100, Its Thumb-Shift Keyboard, and Rival Word Processors
In May 1980, Fujitsu announced the OASYS 100. It featured a unique "thumb-shift" keyboard, which is described as making Japanese input easier [4]. As the name suggests, it was a keyboard layout that could also be operated with the thumbs. Fujitsu was not the only company making dedicated machines. Canon had the "Canoword," and Casio also released dedicated word processors [7][5]. This shows that various companies competed by innovating ways to make Japanese input easier.
6. How Personal Computers Replaced Dedicated Word Processors
The market for dedicated word processors shrank. Casio explained that their market, which had been gradually shrinking since its peak around 1985, accelerated its decline as the personal computer market grew rapidly with the arrival of Windows [5]. Note that the peak year is based on Casio’s explanation and may differ in other sources. There were also moves to make dedicated machine assets usable on personal computers. In 1996, Canon released software that allowed users to use Canoword documents, external characters, and registered words on Windows [7]. In February 2001, Fujitsu announced the end of sales for the OASYS series. New development had stopped the previous year, citing the "shrinking word processor market." The series had sold over 4 million units, and thumb-shift users were guided to transition to personal computers [6]. While some dictionaries state production ended in 2000, the official announcement was in February 2001. On September 5, 2001, Casio announced its withdrawal from dedicated word processors. Production had already stopped in August, and sales would end once inventory ran out [5]. Being "dedicated" was a strength for document creation, but as general-purpose personal computers spread, this advantage may have faded. This is the article’s perspective, not a definitive cause stated by all sources. ITmedia also notes that the shift from paper output to digital distribution was part of the background [8].
7. Try It Yourself: Converting a Tricky Japanese Sentence on a Phone or Computer
On a smartphone or computer, try entering the sentence "kisha no kisha ga kisha de kisha shita" using only hiragana. See if it converts to something like "Your company's reporter returned to the company by train." Next, try breaking it down into chunks like "kisha no" and "kisha ga" and converting them one by one. Do the results differ from converting the whole sentence at once? Look at what candidates appear. Conversion results vary by device and settings, so they may not match the JW-10’s results. For original details on the JW-10’s size, price, and system, read the PC Watch article (in Japanese) [2].