There is little available evidence for claims of the efficiency of alternative keyboard layouts. Part of the difficulty in providing such evidence is the prior experience of research subjects with the standard keyboard layout. The purpose of this study was to provide a true comparison of the QWERTY and Chubon keyboard layouts for individuals who type with a single digit. A single subject, repeated measures design was used with a convenience sample of nine participants. Each participant began typing on a pre-selected keyboard and continued typing until fluency was achieved. This procedure was replicated with each keyboard layout. The words per minute typed at fluency for the ReverseQWERTY was approximately 62% of the QWERTY, indicating that the learned effect had been erased. The average typing speed of the Chubon was at least 5% higher and at most 51% higher than the ReverseQWERTY. There were no significant patterns of error. Results of this study indicate that the biomechanical layout of the Chubon is superior to that of the ReverseQWERTY, and by extension the QWERTY. Future research regarding the topic is needed to further expand knowledge of the effectiveness of the various alternative keyboard layouts.
The QWERTY keyboard, which was named after the first six letters of its layout, is the standard design for both typewriters and computer keyboards. The QWERTY was originally designed to decrease the pace of text entry and to prevent key jamming in early mechanical typewriters. Modern computers to not suffer from key jamming, and many feel that it is time to consider alternative keyboard layouts. Because the fluency of text entry is determined by the speed and accuracy of the user, it is important to examine these measures for alternative keyboard options. One option presently available is the Chubon keyboard, which was designed to improve the speed and efficiency of single digit entry (Chubon & Hester, 1988). Consequently, a comparison of the QWERTY keyboard and the Chubon keyboard can provide valuable information for occupational therapists who prescribe alternative keyboards for persons with disabilities, including those with low endurance and decreased strength.
Computer knowledge and the ability to keyboard efficiently have become essential employment skills for the majority of jobs. No longer are they required only for newspaper reporters and secretaries (Struck, 1999). Access to internet communication, educational computer programs, professional documentation and record keeping would not be as efficient without computer skills. The keyboarding proficiency that is developed in school prepares the student for the working world where such skills are used to enhance productivity (Struck, 1999). Computer related technology is also influential in the potential lifestyle development of people with disabilities (Hurlburt & Ottenbacher, 1992). Access to electronic communications can reduce the barriers in the home, community, and workplace that limit independence and productive employment (Hurlburt & Ottenbacher, 1992). In most automation and computerization, the speed at which data are entered plays a key role in the performance of the system as a whole (Shieh & Lin, 1999). The layout of the keyboard is a crucial variable in determining the speed, accuracy, ease of learning, and efficiency of the interaction between the user and the computer (Hurlburt & Ottenbacher, 1992).
Because the keyboard layout plays such a substantial role with regard to speed and accuracy, it is important to use a keyboard that meets an individual's needs ( Poole , 1995). The standard QWERTY keyboard layout has frequently been criticized for being inefficient (Hurlburt & Ottenbacher, 1992). Dvorak (as cited in Shieh & Lin, 1999) even went as far as to say that "the QWERTY keyboard was crude and violated ergonomic principles" (p.113). The QWERTY was originally developed for the manual typewriter in order to slow down the typist and to prevent the mechanical keys from becoming jammed (Struck, 1999). Because high frequency keys were placed at the far corners of the keyboard to provide the mechanical clearance necessary to avoid key jamming, extreme travel of the fingers occurs. This type of configuration requires full finger movement within their range of motion, possibly placing them at high risk for overuse problems (Struck, 1999). Greater finger travel between letters may also result in increased fatigue (Kincaid, 1999). Those experiencing visual, auditory, cognitive, and language limitations have also reported problems with the QWERTY layout (Fong Lee, 1995). After comparing the QWERTY to an alternative keyboard, subjects reported that while using the QWERTY, they 'worked harder' and had more difficulty remembering the layout of the keys (Hurlburt & Ottenbacher, 1992).
Historically, there were many recommendations for an improved keyboard design soon after the QWERTY keyboard was developed (Shieh & Lin, 1999). Early suggestions for improvement included placing the higher frequency letters in the center of the keyboard on the basis that the index and middle fingers were the fastest (Shieh & Lin, 1999). The 'Simplified Keyboard', proposed by Dvorak and Dealey in 1936, emphasized the need to minimize hand and finger motion and to apply the principles of ergonomics (Shieh & Lin, 1999). In 1988, Robert Chubon developed a new keyboard pattern for the single digit typist (See Figure 1). In the Chubon layout, the most commonly used letters of the alphabet are placed in the center of the keyboard. Those letters which are often paired are also placed in close proximity. Altogether, the Chubon keyboard pattern has been documented to require about 37 percent less finger travel than the QWERTY keyboard for corresponding blocks of text (Anson, 1997). This "frequency-of-use" key layout was created to facilitate text entry and reduce fatigue (Kincaid, 1999). According to previous research, this type of layout increased typing speed by minimizing distance traveled by the fingers (Hurlburt & Ottenbacher, 1992).
Keystroke speed and accuracy have been utilized as performance measures for evaluating various writing/typing systems (Hurlburt & Ottenbacher, 1992). Hurlburt and Ottenbacher (1992) conducted a study which compared the results of these measures between the QWERTY keyboard and the Dvorak keyboard. However, this study did not suggest a significantly strong contrast between the two to support a change of keyboard layout (Hurlburt & Ottenbacher, 1992). One reason for this may be that the subject's prior experience with the QWERTY keyboard affected the results (Hurlburt & Ottenbacher, 1992). This learned-effect of the QWERTY keyboard must be controlled in order to achieve an accurate comparison. If the user's prior knowledge of the QWERTY layout can be removed, a valid comparison between the alternative keyboards would become possible.
In order to control for this prior learning, this study uses a variant of the standard QWERTY layout, developed to maintain the biomechanical relationships of the keys while providing a novel arrangement. This is accomplished by inverting the QWERTY layout vertically and horizontally. A dual axis inversion is considered necessary because Matias, Mackenzie and Buxton (1994, pp.88-94) have demonstrated that right to left inversions are easily learned. The ReverseQWERTY pattern (see Figure 2), used in this study, allows comparison of two novel keyboard patterns, while maintaining the inter-key relationships of the QWERTY keyboard.
This study was intended to test the following research hypotheses:
A single subject, repeated measures design was implemented to test eleven subjects on their typing speed and accuracy for three keyboard patterns: QWERTY, ReverseQWERTY, and Chubon.
A convenience sample was gathered by posting flyers around the campus of a small private college in NE Pennsylvania . The flyers included the purpose of the study and expressed the need for and appreciation of those who participated. The flyer also requested that only those people who are able-bodied, and therefore without perceptual or upper extremity deficiencies, volunteer to participate. The first eleven self identified, able-bodied people, over the age of eighteen, who replied to the flyer and who were ultimately willing to cooperate with and agree to the terms of the experiment were used. As an incentive to participate in the experiment, the subjects were informed that they would receive a ten dollar gift certificate to TGI Fridays for their cooperation.
The source documents for this study consisted of chapters 1-10 of the novel Anne of Green Gables . (Lucy Mongomery) This novel is available in electronic format from Project Guttenburg The text document was formatted in Microsoft Word to remove extraneous carriage returns and to improve readability.
For the purposes of this study, two standard Microsoft compatible personal computers equipped with Microsoft Office 97 were used. Each computer had a Pentium II processor operating at 450 MHz, 128 MB of memory, and a 17 inch monitor. An Intellikeys® keyboard, which is flat surfaced and more upright, replaced the normal keyboard system for all text input. The alternative keyboard patterns were provided using OverlayMaker, which is also available from Intellitools. It is important to note however, that the Intellikeys keyboard is not the only entry system that can be used with the Chubon pattern: it can also be used with the standard keyboard, and with many on-screen keyboards.
All text was entered in to Microsoft Word 97, which also provided measures of speed and accuracy. To determine speed, each individual was timed with a stopwatch, and was asked to stop typing after exactly twenty minutes of typing. The "Word Count" accessory in Microsoft Word provided a count of the number of words completed in the time interval. Words per minute were calculated using the formula:
Words typed in 20 minutes
= Words per minute
Accuracy was determined by how closely the individual's document matched the source document. At the end of a session, the "Compare Documents" feature of Microsoft Word was used to perform a check of the subjects typing. Each deviation identified by "Compare Documents" was considered a single error. Single errors could include the dropping a letter, reversal of letters, or even the deletion of entire lines of text. The degree of error was calculated by dividing the total number of words into the number of errors (DeVries, Deitz, & Anson, 1998).
Participants in the study typed on each of the target keyboards until they achieved the level of 'fluency'. For purposes of this study, fluency was considered to be three consecutive trials with a change in typing speed of less than 7% between adjacent trials. When a subject achieved fluency with one keyboard pattern, they switched to the next target keyboard. Keyboard orders were balanced so that keyboard order did not confound the results.
The test sessions for this study were conducted in an internal room of a college building that housed a computer lab. The internal room minimized external distractors in the course of the study. At each treatment session, the subjects were seated at the Windows compatible computer, with the Intellikeys keyboard pre-programmed to the target keyboard pattern, and an overlay of the keyboard pattern installed. The target text was positioned to the side of the computer monitor preferred by the subject. At each session, the subject received a new section of the target text, so that familiarity would not influence typing speed.
At the word "Go", the subject began typing, and a stopwatch was started by the researcher. After exactly 20 minutes, the researcher signaled "Stop", at which point the subjects saved their typing with identifiers for name, date, and trial number. To avoid excessive fatigue, no more than three trials were conducted in a single session, though some subjects participated in more than one session per day. At the end of each day the results were graphed to determine whether or not the subject had achieved fluency with the current keyboard. This process was continued until the subjects had achieved fluency with each target keyboard.
At the end of each session, the subjects' output was compared with the source text using Microsoft Words "Compare Documents" feature. The number of deviations between the source and the typed text was recorded as the number of errors for that session. The "Word Count" feature was used to provide the words-per-minute fluency for each session. The words per minute were recorded, and the difference between each session and the prior session was calculated to determine the percent change in typing speed. The subjects continued with each keyboard until three consecutive sessions showed typing speeds within 7%.
Typing speeds and errors were plotted as line graphs after each successive trial. Over successive trials, these graphs showed improvement in typing speed to plateau, and increasing or decreasing accuracy. Fluency was observable when the celeration line displayed a plateau, or when the typing speeds of three consecutive trials were within seven percent of each other (Krishef, 1991).
Of the eleven participants that began, only nine completed the study. Those subjects who did not finish the study each completed one keyboard layout before electing not to continue. Five of the nine participants who did complete the study achieved fluency on two keyboard layouts. The remaining four participants achieved fluency on all three.
The results of this study support the assertion that the ReverseQWERTY keyboard pattern removes the effect of learning while maintaining biomechanical relationships between the keys.
Six subjects achieved fluency on both the QWERTY and ReverseQWERTY keyboards. All of the subjects typed slower on the ReverseQWERTY keyboard, with an typical typing speed 62% slower on ReverseQWERTY than QWERTY. Since the two keyboards are mirror images, this difference cannot be due to mechanical difficulty, and must be the effect of experience on the QWERTY keyboard. (It is of interest that this finding matches closely with the empirical observation of the first author that experience produces an improvement of approximately 50% in text generation speeds over the first plateau.). Thus, the ReverseQWERTY layout may be considered to provide, for single digit typists, a fair comparison of a novice typist using the QWERTY keyboard.
Seven subjects achieved fluency on both the Chubon and ReverseQWERTY keyboard patterns. In each case, the typing speed of the Chubon was at least 5% higher than that of the ReverseQWERTY, and was as much as 51% higher. Typically, the initial speed of text entry with the Chubon was equal to or greater than the final speed of the ReverseQWERTY. This supports the second hypothesis, that the biomechanical layout of the Chubon is superior to that of the ReverseQWERTY, and by extension, to the QWERTY keyboard pattern, for single-digit typing.
There were no significant patterns of error related to the keyboard layout or to the number of typed words per minute. The scales of error varied among the individual participants. However, each of the participants remained within an average of 2% of their own error rate with each keyboard layout. This suggests that the degree of error is not specific to the style of the keyboard. This finding does not support the hypothesis that the Chubon pattern provides more accurate typing than QWERTY. Similarly, the number of trials required by each individual to achieve fluency between keyboards was not significantly related, suggesting that this also was not dependent upon the keyboard type. The typical results of the study are as shown in Tables 3, 4, and 5.
There are numerous conditions that may limit a person to single-handed, or even single-digit typing. For instance, a person with a C5-C6 spinal cord injury, amputation, or hemiplegia secondary to a cerebral vascular accident, may have partial to no functioning with one hand (Pedretti, 1996). As occupational therapists, it is our task to adapt the environment to meet the individual needs of these clients. Whether for functional communication, leisure, or work, the effective use a computer and the Internet is a growing need for the disabled population. In order to provide maximum facility in typing, the keyboard layout should be matched to the physical skills of the typist. The Chubon keyboard provides a pattern that supports typing with a single digit on the physical keyboard, or with the mouse as an on-screen keyboard.
Unfortunately, and possibly due to a lack of knowledge, there is a reluctance to learn, and ultimately to change keyboard layouts. Some individuals do not want to appear "different" from their peers, and so resist alternative patterns. Others, on first experience, note that they type slower on an alternative than on the familiar keyboard, and abandon the recommended change before the benefits can be realized. This study offers solid evidence that the biomechanical layout of the Chubon is superior to that of the QWERTY. When prescribing a keyboard layout to a client who will be typing with a single digit, the Chubon keyboard should ultimately provide for more efficient and faster typing.
The therapist must keep in mind the fact that, after a neurological insult, some clients will display limitations in new learning. In these cases, the advantages of relying on old learning (the QWERTY pattern) may outweigh the benefit of hard-won new learning, and a new keyboard pattern may not be indicated. But for those clients with the ability to learn a new pattern, or for those without significant prior learning (as in school system practice), the performance benefits of the Chubon pattern support the effort to use this non-standard pattern.
Currently, operating systems such as Windows 98 and Mac OS 9 include keyboard patterns for one-handed typing as options in the installation media. They do not include single digit keyboard patterns because of the paucity of evidence for the benefit of these patterns. This study provides evidence that may support the inclusion of the Chubon pattern in future releases of computer operating systems so that it can be installed without any external software. But even today, public domain programs that allow keyboard remapping are available from many on-line sources. These programs make it possible to install the Chubon pattern on any computer. Some of these programs do not require that software be added to the host computer, so may be used without danger on public computers. It is very easy to create and add the Chubon pattern to a computer, and, for the individual who types with a single digit, the potential benefits are substantial.
Future research regarding this topic is desperately needed to further expand knowledge of the effectiveness of the various alternative keyboard layouts. Because this study has shown that the ReverseQWERTY removes the learned effect of the original QWERTY, future researchers may use this methodology to compare other keyboard patterns for special populations to the standard. When these other layouts are intended for single digit typists, they may be compared with the Chubon as well, with the knowledge that the biomechanical advantages of the Chubon as compared with the QWERTY layouts have been established. Further research conducted with the Chubon keyboard layout may indicate maximum attainable speeds, which can in turn be compared with the plateau speed of the QWERTY for single digit typists. Having equivalent knowledge of both keyboard layouts will make it possible to demonstrate a true comparison of the Chubon and the original QWERTY. To increase generalizabilty, this study should be repeated with people with functional limitations as subjects.
Finally, the methodology of this study provides a technique for comparing alternative text generation systems. Additional studies comparing other input methods, such as physical keyboards versus on-screen keyboards, using the same methodology, might generate a body of knowledge that would allow informed selection of alternative input technologies for people with disabilities.
Confounding variables are elements that are not the subject to manipulation, but which might provide systemic differences within a study. The only identified variable that could confound this study is past experience. For example, if a subject has had extensive experience with the original QWERTY, they may have found it more difficult to assimilate to its reversed counterpart. It is possible that the cognitive dissonance associated with the distortion of the QWERTY placed the ReverseQWERTY pattern at a disadvantage compared with the completely novel Chubon pattern. However, considering the similarity in time to achieve plateau and in error rates between the keyboard patterns it is unlikely that any such effect was operating.
Limitations for this experiment and its potential results may include the use of a purely able-bodied sample, which may limit generalizability to the disabled population. This population was selected, however, to separate the effects of disability from the effects of keyboard pattern. A replication of the study using people with disabilities may show interaction effects that either weaken or strengthen this study.
Anson, D. (1997). Alternative computer access: A guide to selection. Philadelphia : F.A. Davis.
Bailey, D.M. (1997). Research for the health professional: A practical guide (2 nd ed.). Philadelphia : F.A. Davis.
Chubon, R.A. and Hester, M.R. (1988). An enhanced standard computer keyboard system for single-finger and typing-stick typing . Rehabilitation Research and Development, 25, 17.
DeVries, R.C., Deitz, J., Anson, D. (1998). A comparison of two computer access systems for functional text entry. American Journal of Occupational Therapy, 52 , 656-665.
Fong Lee, D. (1995). Alternative keyboards. Canadian Journal of Occupational Therapy, 62 , 175.
Hurlburt, M. & Ottenbacher, K.J. (1992). Spinal cord: Wounds and injuries. Journal of Rehabilitation Research and Development, 29 , 54-64.
Kincaid, C. (1999). Alternative keyboards. Exceptional Parent, 2, 34-35.
Krishef, C. H. (1991). Fundamental approaches to single subject design and analysis. Malabar , FL : Krieger.
Matias, E., MacKenzie, I. S., & Buxton, W. (1994). Half-QWERTY: A one-handed keyboard facilitating skill transfer from QWERTY. Proceedings of the INTERCHI '93 Conference on Human Factors in Computing Systems (pp. 88-94). New York : ACM.
Pedretti, L.W. (1996). Occupational therapy: Practice skills for physical dysfunction . St. Louis , Missouri : Mosby.
Poole , C.J.M. (1995). Computers and the handicapped. British Medical Journal, 311 , 1149-1152.
Shieh, K.K. & Lin, C.C. (1999). A quantitative model for designing keyboard layout. Perceptual and Motor Skills, 88 , 113-125.
Struck, M. (1999). Focus on. One handed keyboarding options. OT Practice, 4 , 55-56.