Human Factors and Linguistic Considerations: Keys to High-Speed Chinese Character Input

HUMAN FACTORS AND LINGUISTIC CONSIDERATIONS: KEYS TO HIGH-SPEED CHINESE CHARACTER INPUT Paul L. King" Cornell University Ithaca, NY, USA Abstract With a keyboard and supporting system developed at Cornell Univer- sity, input methods used to identify ideographs are adaptations of well- known schemes; innovation is in the addition of automatic machine selec- tion of ambiguously identified char- acters. The unique feature of the Cornell design is that a certain amount of intelligence has been built into the machine. This allows an operator to take advantage of the fact that about 60% of Chinese characters in text are paired with other characters to form two-syllable compounds or phrase words. In speech and writing these pairings eliminate about 95% of the ambiguities created by ambiguously identified syllables. Introduction For Chinese character input to computers, a Cornell research team has approached the problem from the point of view of the type of person who would most likely be operating a Chinese electronic typewriter, namely, a commercial Chinese typist with junior middle school education who would regularly be typing for eight to ten hours per day. For such a person, a word processing system should be easy to learn, fast (aver- aging 50 characters/minute), and cap- able of being used for several hours without inducing a high level of fa- tigue. Of the many input systems that have been proposed in recent years, one based on Wang Yun-yu's four-corner numbering system has best demonstrat- ed, in the opinion of the Cornell team, the capability of meeting the criteria of ease of learning, speed, and low operator fatigue level. For example, operator training is simpli- fied because keystrokes are assigned by "inspection" rather than rote memory. Also, frequently used simplex characters such as particles are given unique identifiers so they can be inserted in text without going through a manual disambiguation process. Even more significantly, manual disambiguation has been eliminated entirely in nineteen cases out of twenty by attention to ling- uistic affinities of characters. Shape Code The four-corner system is a simple encoding scheme that native Chinese speakers learn in about half an hour. In this system, the peri- pheral forms of all Chinese characters are projected onto ten basic stroke shape classes (~,--,~,~,~ ,~ , ~ , 7,/~,~), to which single-digit values are assigned. On the basis that ideograms are basically square in appearance, four-digit numbers can be read from stroke shapes at the corners, in the sequence top left, top right, bottom left, bottom right. Thus the shape classes describing a character such as ~ are.~,J~,~,--. I have made two refinements on Wang's original four-corner system. First of all, the Cornell code in- cludes elimination of all four-corner null zeros, so that identifiers for characters vary from one to four digits. Thus, the identifier for the Chinese character "--" is simply "~", with the three null zeros eliminated. As applied to the Cornell input system, middle-of-word ("comma") and end~of-word ("print") delimiter keys make possible the use of var- iable-length input codes for …

Content truncated to 500 words