An Assignment of Key-Codes for a Japanese Character Keyboard

Yuzuru Hiraga, Yoshihiko Ono, Hisao YamadaCOLING 1980

TL;DR

This paper proposes a novel multi-stroke input method and key-code assignment for Japanese characters, aiming to achieve touch-typing efficiency comparable to English keyboards. By focusing on optimizing finger movements and minimizing awkward sequences, the authors demonstrate a system that could significantly improve Japanese text processing speed and reduce typist fatigue.

What Problem Does This Paper Address?

The advent of computers revolutionized text processing in Western societies, largely due to the efficient input provided by the QWERTY keyboard. However, Japan and other non-alphabetical countries have lagged behind in developing similarly efficient input methods for their complex writing systems. The primary challenge lies in the vast number of characters (thousands of Kanji, plus Kana and numerals) required for Japanese text. Traditional one-to-one key mapping is impractical due to the sheer number of keys needed. Existing multi-stroke methods often rely on mnemonics, which can hinder the development of subconscious, error-free touch-typing. This paper aims to bridge this gap by designing a key-code system specifically for touch-typing Japanese characters, prioritizing speed, reduced fatigue, and accuracy.

Methods

The core of the proposed solution is a multi-stroke, 2-levelled code system designed for a compact keyboard (30-40 keys).

  • Basic Character Set: Approximately 900-1200 frequently used characters are assigned 2-stroke codes. The assignment prioritizes efficiency of finger movements over mnemonics. This involves:
  • Rhythm and Alternate Hand Stroking: Maximizing the use of alternating hands (e.g., Right-Left-Right-Left) to maintain a steady typing rhythm.
  • Row Preference: Minimizing hand movement between rows by favoring the home row.
  • Finger Dexterity: Distributing the load proportionally to finger strength and dexterity.
  • Minimizing Awkward Sequences: Assigning negative weights to undesirable key combinations (e.g., "hurdling" over rows, "reaching" with the same finger).
  • Outside Character Set: Less frequent characters (around 2000) are assigned longer codes (6 strokes). These codes utilize a 2-stroke entry code followed by 2-stroke codes derived from component characters (radicals or parts of Kanji). This approach allows for mnemonic recall without compromising the efficiency of the basic character set.
  • Performance Simulation: The authors employed computer simulations to model typing motion and evaluate the efficiency of their proposed code system against the QWERTY keyboard.

Datasets

The paper mentions the use of:

  • Newspaper Text Data: Used to determine character frequencies for selecting the basic character set.
  • "Text Data from Three major Japanese Newspapers" (1966): This dataset, containing 1,500,000 characters, was used for statistical analysis of character usage.

No new benchmarks were introduced; rather, the authors compared their system's performance against existing data for the QWERTY keyboard and the Dvorak Simplified Keyboard (DSK).

Key Insights and Ideas

  • Touch-typing is Achievable for Japanese: The paper strongly advocates that touch-typing, with its inherent efficiency and reduced cognitive load, is not only possible but desirable for Japanese text input, contrary to common assumptions.
  • Efficiency Over Mnemonics for Basic Characters: For frequently used characters, prioritizing subconscious, conditioned reflexes through optimized key assignments is more effective than relying on conscious mnemonic recall, which can interrupt typing flow.
  • A Two-Tiered Approach: Separating frequently used characters (basic set) from less frequent ones (outside set) allows for distinct optimization strategies, balancing pure efficiency with mnemonic support where it's most practical.
  • The Importance of "Awkward Sequence" Minimization: The authors found that reducing specific types of awkward key combinations significantly impacts overall typing efficiency and reduces errors.
  • QWERTY's Suboptimal Design: The analysis highlights that even the standard QWERTY keyboard, while functional, has significant inefficiencies in terms of hand and finger distribution and awkward sequences, which their proposed Japanese system aims to surpass.

Key Results

The paper presents quantitative results from simulations and analyses:

  • Hand Distribution: The proposed "T-code" system achieved a more balanced distribution between left and right hands (47.5% vs. 52.5%) compared to QWERTY (57.2% vs. 42.8%).
  • Row Distribution: T-code significantly favored the home row (58.5%) compared to QWERTY (31.8%), indicating less hand movement.
  • Finger Distribution: T-code showed a more even distribution across fingers, with QWERTY showing a notable overload on the right index finger.
  • Rate of Awkward Sequences: T-code achieved a much lower rate of awkward sequences (2.1% for Hurdle, 5.2% for Reach) compared to QWERTY (9.6% for Hurdle, 8.2% for Reach).
  • LR-Sequence Length: The expected length of alternate hand stroking sequences for T-code was significantly higher (7.69) than for QWERTY (1.09), indicating superior alternate hand usage.
  • Simulation Results: Computer simulations suggested that T-code requires 20-100% less hand motion than QWERTY, and its performance is comparable to the Dvorak Simplified Keyboard (DSK) for English.

Limitations and Future Work

  • Experimental Validation: The authors acknowledge that the ultimate test would be actual user experiments and training to measure productivity and mental fatigue objectively. They have not yet conducted these experiments.
  • Objective Measurement of Fatigue: Measuring mental fatigue objectively remains a challenge.
  • Outside Character Coding: While a method for coding outside characters is proposed, the authors note that the 6-stroke length might seem high, though its infrequent appearance in text mitigates its impact on overall speed. They also mention the possibility of conflicts where a typed code might not be for the intended character, requiring typist verification.
  • Further Refinement of Codes: The authors suggest that secondary modifications to the codes, considering character pair distributions in Japanese texts, could further improve performance, though they believe significant reductions in awkward sequence rates are unlikely.
  • Training Methodology: Developing an appropriate and efficient training method is identified as a crucial task for widespread adoption.

Summary updated

BibTeX
@inproceedings{hiraga1980assignment,
  author = {Yuzuru Hiraga and Yoshihiko Ono and Hisao Yamada},
  title = {{An Assignment of Key-Codes for a Japanese Character Keyboard}},
  booktitle = {International Conference on Computational Linguistics},
  year = {1980},
}

Loading related content...

Figure