1. Why One Key per Kanji Does Not Work for Japanese Typewriters
For an English typewriter, 26 alphabet letters plus numbers and symbols are enough. Japanese is not so simple. According to the IEEE Milestone description, standard Japanese characters include about 48 kana and more than 6,000 kanji. Writing a normal sentence requires both of these sets. [2]
If you divide 6,000 by 48, the result is roughly 125 times more characters. The kana used for input fit in the palm of your hand, but if you assign one key to each kanji, they would not fit even if spread across a whole desk. [2]
This difference in numbers made mechanical input for Japanese difficult. The first solution was not to add more keys, but to arrange movable type on a flat tray and have the user "search and strike" them. [2]
2. How Sugimoto Kyōta's Japanese Typewriter Picked Up Type with One Bar
Sugimoto Kyōta created the Japanese typewriter. He completed a prototype in 1914, obtained a patent in 1915, and the first commercial model was sold by the Nihon Shojiki Shokai in 1916. [1]
Based on materials from the Patent Office history, it is said that Sugimoto studied the frequency of characters used in official documents and selected about 2,400 characters. [7] He chose only the frequently used characters from the thousands of available kanji. However, some explanations state that the number of characters on the tray was about 3,000, suggesting there may have been differences depending on the model or period. [3]
The movable type is packed tightly on a flat tray. In the center are three types of kana: hiragana, katakana, and Gothic-style kana, with kanji placed around them. [3]
The typist moves a single type bar forward, backward, left, and right to position it over the desired character. The bar picks up that type and strikes it onto paper wrapped around a cylinder. [7]
In other words, instead of adding more keys, the system arranged characters by their "storage location," relying on the user’s eyes and hands to find them. [7]
3. How Searching for Type Slowed Down Typing Speed
Explanations on Kotobank list the number of characters printed per hour by font style, such as 1,100 characters for regular script and 1,300 for running script. [1] This shows that typing speed varied clearly depending on the font and machine model. [1]
Searching itself was the work. On Japanese typewriters, operators had to find and print each required type one by one. [1] People who memorized the arrangement of thousands of characters could type quickly. This skill was required in office settings. [1]
A machine from around 1935 is held in the Postal Museum. [8] By the early 1940s, it had become a representative office machine. [1] However, the page had to be retyped when text was corrected. [1]
4. Why Toshiba Began Researching Typing in Kana and Converting to Kanji
In 1971, research began at Toshiba on converting characters typed in kana into kanji. [2] The research team was led by Kenichi Mori. [2]
The idea can be summarized simply: stop "searching and striking kanji" and instead "type the reading and let the machine choose the kanji." Since only kana is used for input, about 48 characters are sufficient. [2]
However, this introduced a different difficulty. Japanese has many words with the same reading but different kanji. If you type "hashi," you must determine from the context whether it means bridge or chopsticks. [2] IEEE materials state that choosing the correct character from thousands of kanji and kana was the key to this technology. [2]
The job of searching shifted from the person to the machine. So, how did the machine choose? [2]
5. How a Built-In Dictionary Let the Machine Choose Kanji
According to a PC Watch article, the JW-10 pre-registered 54,000 common words and 8,000 proper nouns in its dictionary. [4] This totals about 62,000 words. If the kana typed matches a word in the dictionary, the machine can display the corresponding kanji candidates. [4]
Toshiba built a prototype with conversion features in August 1977 and announced the completed automatic conversion system in September 1978. It was displayed at Data Show the following month, and the first product, the JW-10, was shipped in February 1979. [2] Some sources list the release as March, so this article distinguishes between the announcement in September 1978 and the shipment in February 1979. [2]
In evaluations introduced by Gendai Business, it was noted that even non-specialists could input Japanese quickly. Furthermore, the ability to have the machine remember text and make corrections or insertions was a function impossible with typewriters. [5]
Without the skill to search through thousands of characters, you could create text just by typing hiragana. The entry point for input had widened significantly. [5]
6. Why Falling Word Processor Prices Made Typewriters Disappear
According to Wikipedia, the average price of Japanese word processors fell from about 2 million yen in 1980 to about 164,000 yen in 1985. [6] Dividing 200 by 16.4 gives approximately 12, meaning the price dropped to about 1/12th in five years. [6]
There is an explanation that the term "word processor" became widespread through a Fujitsu commercial in 1982. [6] As prices fell, these machines spread not only to companies but also to homes. [6]
According to IEEE materials, production of dedicated word processors reached 2.71 million units per year in 1989. [2] By 1999, personal computer sales are said to have surpassed dedicated machines. [2]
In their shadow, Japanese typewriters disappeared from the 1980s onward. [1] The machine that searched through a tray of thousands of characters passed its role to the machine that used kana input. [1]
The era of dedicated machines was short, but the input system is evaluated as having been passed down to kanji input on personal computers and mobile phones. [5]
7. How to Check Conversion Candidates on Your Own Smartphone
Try typing the hiragana "hashi" on a smartphone or computer at home. How many conversion candidates appear? [4] Try "kisha," "kouen," and "ishi" as well, and compare the number of candidates. [4]
You may find readings with many candidates and others with few. [4] This difference is the new form of the job of "choosing," which was born after the machine took over the job of "searching from thousands of characters." [4]
You can read the dictionary word counts in the PC Watch article in Source [4]. The price figures are in Source [6]. Both are in Japanese. [4]