Keyboard Research 
Abstracts & Summaries 


Prev ] Home ] Up ] Next ]

Alden, Daniels, & Kanarick, 1972

A search of the psychological, technical, and promotional literature was conducted to compile information relevant to key, keyboard, and operator characteristics. The most recent and significant articles were discussed and evaluated. Where possible, general conclusions have been drawn to aid the keyboard designer.

A review of the recent literature indicates that much research effort has been expended on the operator-keyboard interface. However, as the following statements indicate, there are few definitive findings on which to base design standards. The following are general design considerations as abstracted from the discussed reports.

Key diameters of 0.5 in. (1.27 cm.) and 0.75 in. (1.81 cm.) center-to-center spacings are most typical.

Force and displacement. Ranges of 0.9 to 5.3 oz. (25.5 to 150.3 g.) force and 0.05 to 0.25 in. (0.13 to 0.64 cm.) displacement are preferred by operators, although force and displacement have slight effect on keying performance.

Visual, auditory, or kinesthetic feedback shows little effect on performance by experienced operators, but visual feedback appears to be important during training.

Miniaturized keyboards used for touch operation show performance decrements, but size itself has little effect on self-paced or sequential operation.

Ten to 35 degrees seems an acceptable range for inclination of the working surface. However, this has questionable influence on operator speed and accuracy, even though most typists preferred some keyboard inclination.

A top-to-bottom, left-to-right cluster of numbered keys (touch-telephone format) seems to be the superior arrangement from the available evidence.

Operator preferences do not correlate highly with measured performance.

Chorded entry may be faster than sequential entry, especially for some special tasks, such as mail sorting.

Although research on keyboard design and use continues, the sponsorship and motivation for such work often precludes the systematic study of any single keyboard parameter. The majority of the current research findings in the open literature have been undertaken to evaluate specific product lines. It is very seldom possible, therefore, to obtain objective data on the relative speed or accuracy of performance across various generic types of keyboard designs. This situation results in a proliferation of keyboard and task-specific data. Unfortunately, such data are often incompletely reported or involve unique metrics, making interstudy comparisons difficult or impossible. Consequently, the major issues of today are the same as those of a decade or more ago. Furthermore, until there is a systematic effort to produce relating, unambiguous, quantitative data on the parametric relationships between keyboard design and operator performance, such issues will remain unanswered.


Armstrong, Foulke, & Martin, 1994

This paper considers one way that occupational health professionals can assess the force exerted by keyboard users and the possible relationship between that force and the key force-displacement relationship. First, three personal-computer keyboards with the standard QWERTY layouts were tested as described by the American National Standard for Human Factors Engineering of Visual Display workstations (ANSI/HFS 100-1988) to determine the peak forces, 0.47-0.89N; displacements prior to the "breakaway" force that acknowledges key registration, 2.0-2.5 mm; and total key travel, 3.3-4.3 mm. Second, keyboard reaction forces were recorded while 10 subjects typed 4 alphanumeric sentences on the keyboards. It was found that the peak forces corresponding to each keystroke were 2.5 to 3.9 times the required activation forces, indicating that the subjects consistently displaced the keys to their limits. The average of the peak forces for all keystrokes was lowest for the keyboard with the lowest required activation force. It was concluded that keyboard reaction forces can be used as an index of finger forces for keying tasks. Further studies are necessary to evaluate the relationship between keyboard reaction forces, fatigue, and chronic muscle, tendon, and nerve disorders.


Bailey, 1982

Bailey briefly mentions the work of Klockenberg and Kroemer: In 1926 a German named Klockenberg published a book dealing with the design and operation of the typewriter. He described how the keyboard layout required the typist to assume postures of the trunk, head, shoulders, arms, and hands that were unnatural, uncomfortable, and fatiguing. Klockenberg suggested a number of improvements to the keyboard layout, some of which are still valid. Probably his most interesting idea was to separate the keyboard sections allotted to the left and right hands to alleviate tension in the typist’s shoulders and arms. Kroemer (1972) published a study showing that this layout produced better performance. (Bailey, 1982, p.301)


Brigham. & Clark, 1986

The purpose of this study was to use the ISO Draft Proposal Standard for the Assessment of the Ease of Use of Alphanumeric Keyboards (TC 159/SC4) to determine if the STR ergonomic keyboard (the commercial version of the keyboard developed by Nakaseko et al., 1985) was ISO-compliant (equal performance and preference as for a standard, know-to-be-ISO-compliant keyboard). Twenty experienced typists participated in the study. They had 20 minutes of practice with each keyboard, then types for seven 20-minute sessions separated by 5-minute rest periods. The authors report: "The performance on the standard keyboard is clearly better than that on the Ergonomic keyboard after 2.5 hours use. Not one subject had a better performance on the Ergonomic keyboard than they had on the standard one. However, . . . after 2.5 hours, performance on the Ergonomic keyboard continued to improve while on the standard one it appeared to level out . . . What is perhaps surprising is the amount of learning that has taken place on the standard keyboard.

At the end of the study, participants completed questionnaires that included:

Which keyboard did you find the most comfortable to use?

STR: 3, Standard: 16, No difference: 1

You may have found the split keyboard very awkward at first. Having used it for a while now, how would you compare your performance to using a normal keyboard?

Better: 0, Worse: 17, Same: 3

Which type of keyboard would you prefer to use yourself?

STR: 1, Standard: 18, No preference: 1


Buesen, 1984

The first truly ‘ergonomic’ keyboard for modern VDU’s was shown to the public in April 1983 in Zurich. This split keyboard was the result of a research program carried out by the Department of Hygiene and Work Physiology, ETH, Zurich. The ergonomic factors of the keyboard were convincing to the press, computer manufacturers and users. The steps from models of this keyboard to the final product and related problems of industrial design and engineering are explained. This will help in understanding the distinctive time lags between the results of scientific research and the availability of a product in the market-place.


Creamer and Trumbo, 1960

Five male subjects were given 3-minute trials at each of five keyboard positions for 20 consecutive days. The keyboard, consisting of the eight keys of the starting position of a typewriter, was hinged in the middle, so that the direction of tapping movements could be varied from horizontal to vertical. The task was a simple alternation of both fingers and hands. Rate of tapping was found to be greatest at the positions intermediate between horizontal and vertical keyboards. Errors decreased monotonically from the horizontal to the vertical keyboard positions, but were highly infrequent at all positions after 5 days of practice. No significant differences were found among the experimental conditions on a performance-decrement index of fatigue. Practical implications of the findings were discussed.

Probably the most significant finding of the study from a practical point of view was the relatively poor performance with the standard horizontal keyboard. This keyboard and the experimental task closely resemble the typing task and other multifinger tapping or keying operations. The implications of the study for the design of such keyboards are quite clear. Further research with typists and with additional tasks is needed before practical gains of such modifications can be quantified, however.

The modified typewriter was on a desk 31 inches above the floor. the keyboard inclination angles were 0, 22, 44, 66, and 88 degrees, with 0 degrees corresponding to the horizontal inclination of a standard keyboard. The participants were non-typists, recruited from a general psychology class and paid for their participation. F-tests for tapping rates were significant for both the first and last (fourth) blocks of trials, although none of the post-hoc means comparisons (Student-Newman-Keuls test) were significant. Participants made few errors after five days of practice, but an F-test on errors in the first block of trials was significant. Again, the post-hoc comparison of means was not significant. The results seemed to favor the intermediate angles, with the poorest performance at 0 degrees. All participants reported that the 0-degree setting was the most fatiguing, and split their preferences between the 44 and 66 degree settings.


Douglas, S.D., and Happ, A.J., 1993

Nine employees of a temporary agency participated in the study. They used three keyboards, on standard IBM PS/2 keyboard (IBM Enhanced Keyboard) and two commercially available split keyboards (referred to as the AR and AD keyboards).

Participants started with training the day before the experimental sessions began. During training, they saw demonstrations of and used all three keyboards. Testing took place over three days. Each day, participants typed at each keyboard for one hours from one of three provided texts. The experimenters counterbalanced the keyboard orders, and participants used all three texts each day. Before the second and third sessions each day, participants had a ten-minute warm-up period to help offset carryover effects. Every 20 minutes during a typing session, participants stopped typing and indicated on a body map any areas of discomfort, using a Borg ratio scale to quantify the discomfort intensity for each region. Mandatory 30-minute rest breaks followed each typing session.

The results indicated that participants typed significantly faster (roughly twice as fast on the average) with the standard keyboard than either of the split keyboards. Typists made significantly fewer errors with the standard keyboard. The AD keyboard had about ten times as many errors as the standard, and the AR keyboard had, on the average, about seven times as many. There were no significant differences in discomfort ratings across the three keyboards. The main effect of session was not significant. For overall satisfaction ratings, participants significantly preferred the standard keyboard.

The results of this study were similar to Brigham and Clark (1986), in that participants performed more poorly with the split keyboards and strongly preferred the standard keyboard, at least as an initial reaction. The main effect of session was not significant, so these data do no suggest any tendency for typing performance to improve or degrade as function of keyboard experience.


Duncan, & Ferguson, 1974

A trial of physiotherapy for muscle incoordination and aching (occupational cramp and myalgia) in teleprinter operators revealed an association between these types of symptoms of operating difficulty and disadvantageous operating postures, which in turn were thought to be related to keyboard layout. The purpose of this paper is to analyze the relationship between posture and sumptuous. Subject of symptoms and unaffected operators were interviewed, tested in various ways, and observed in teleprinter operating. Adverse operating postures of arm and hand were, with the exception of two types of posture, more often right than left sided (when they were not bilateral) and commoner in subject than controls. Every part of the upper limb which was a site of symptoms in operating was more often affected on the right side than the left (when not bilaterally affected). Part of limb affected was usually associated with some adverse operating posture of that region. It was concluded that keyboard design and work height predispose to operating postures which in some operators give rise to symptoms in operating.

Summary: The authors made detailed observations of the operating postures of 90 male subject who complained of symptoms in operating teleprinter keyboards. They also observed 45 unaffected telegraphists of corresponding sex, age, length of service and status. The observed operating postures included shoulder abduction (elbows out), shoulder flexion (elbows forward), ulnar deviation (rotated wrists), wrist extension (wrists bent back) and finger extension. The authors included photographs of the teleprinter workstations, which showed no area for telegraphists to rest their wrists or arms while typing. After analyzing differences in posture between affected and unaffected telegraphists, they concluded that shoulder abduction, ulnar deviation, and wrist extension were important predictors of operating symptoms. "The findings establish that there is a relationship between symptoms incurred in teleprinter operations and certain postures adopted in operating. The findings support the conclusions of these authors that keyboard design and work height lead to ill-advantaged postures, repeated adoption of which gives rise in some operators to recurrent symptoms of incoordination and muscle pain."


Ferguson, D., and Duncan, J., 1974

A previous study of muscle incoordination and aching (occupational cramp and myalgia) in telegraphists revealed that the layout of the standard QWERTY keyboard conferred an uneven distribution of load to the fingers. It appeared that layout also affected operating posture, shown to be linked with the occurrence of these symptoms in operating. The aims of the present study were to explore effects of design on the operator, and to examine how layout could be changed to obviate postulated adverse effects. Affected and unaffected telegraphists were interviewed and tested in various ways, and were observed and photographed while operating teleprinters. The load on each finger of the telegraphist and typists conferred by the standard keyboard was analyzed and found to be maldistributed, the ring and little fingers being overloaded. Other features of keyboard design and work layout were identified as likely sources of inefficient operating postures and thus probably of symptoms in operating. An alternative keyboard layout is suggested which overcomes maldistribution, and to some extent the probable source of postural difficulty. The results of the study support other changes to keyboard design suggested by Klockenberg nearly 50 years ago.

Summary: This work was a continuation of Duncan and Ferguson (1974), with an emphasis on the distribution of work across telegraphists’ fingers. On the basis of this analysis, the authors suggested a new assignment of letters to keys, and a rearrangement of key locations to reduce ulnar deviation. They referenced Klockenberg (1926) and Kroemer (1972), and stated that their data supported the ergonomic value of a split keyboard.


Fernstrom, Ericson, & Malker, 1994

This study investigated how ergonomic design influences neck-and-shoulder muscle strain, through keyboard assessment. Muscular activity was measured electromyographically (EMG) from six muscles in the forearm and shoulders of eight experienced typists using each of five different types of keyboard: one mechanical, one electromechanical, and one electronic typewriter; one personal computer/word processor (PC-XT) keyboard; and one angled at 20 degree in the horizontal plane. The impact on muscular activity of using a palmrest was also studied. The mechanical typewriter induced a higher strain in the forearm and finger muscles than did the modern typewriters and keyboards. These induced no different strain on the neck-and-shoulder muscles, except for the right shoulder muscle, which was more active with the electronic typewriter than with the other machines. Using a palmrest did not decrease the strain on the muscles investigated. Use of the 'angled' PC-XT keyboard did not influence the measured muscular load on the forearm and finger muscles compared to typing on an ordinary PC-XT keyboard, but decreased the extensor muscular strain compared to the electronic typewriter.


Gerard, Jones, Smith, Thomas, & Wang, 1994

EMG was used in evaluation of the Kinesis ergonomic keyboard (Gerard, Jones, Smith, Thomas, & Wang, 1994) which has its keys split apart and set in two concave "dishes" on each side of the keyboard. They determined that the resting posture of the hands on the Kinesis keyboard required significantly less activity to maintain as compared to a standard keyboard. Reduced muscular activity was also measured for typing, which may be attributable to the Kinesis’ unique key arrangement.

Part of a special issue on ergonomics in telecommunications. A study of the electromyographic activity of typists using the Kinesis Ergonomic Computer Keyboard. Six professional typists, aged between 29 and 52 years old, were trained in the use of the new keyboard. Following an initial learning period, electromyographic activity in 4 muscles in the typists' forearms was recorded while the typists were using the keyboard. These data were then compared to measurements taken while the typists were using a standard 101 key IBM PS/2 keyboard. Results showed that, for the flexor carpi ulnaris and the flexor digitorum sublimis, the resting posture on the Kinesis Ergonomic Computer Keyboard required considerably less activity to maintain than the resting posture on the standard keyboard. Moreover, typing with the new keyboard required less muscular activity in the flexor carpi ulnaris, the extensor digitorum communis, and the flexor digitorum sublimis.


Grant, A.H., 1990

This article briefly describes Grant’s split keyboard (the mnemonic ikon keyboard, or MIKey). He references Nakaseko et al. (1985) to justify his design, but does not report any experimentation with the keyboard, The integrated workstation he proposes has the keyboard near the bottom of the display, with no place for a typist to rest wrists or forearms.


Hedge, McCrobie, Land, Morimoto, & Rodriguez, 1995

Hedge et al. (1995) measured chair and computer worksurface dimensions. Ulnar/radial deviation and flexion/extension of the right wrist were recorded dynamically using the Exos Gripmaster System. This system consists of an individually calibrated exoskeleton that fits over the hand/wrist region and simultaneously records horizontal and vertical wrist movements. Wrist movement data, both for individuals and for groups of subjects, were visualized in 3-D using IBM Visualization Data Explorer running on one of Cornell’s supercomputers. A "neutral zone" of movement was determined by using data from Rempel and Horie (1994). Seated posture was assessed using the RULA method (McAtamney and Corlett, 1993). Self-report questionnaires were used to measure musculoskeletal discomfort, work characteristics, personal details, and reactions to the preset tiltdown (PT) system.


Hobday, 1988

A new computer or office machine keyboard has been developed to match hand movements and reduce the postural stress imposed on operators by the Sholes (qwerty) design which engenders fatigue and can lead to pain and disability. The qwerty letter layout has been retained for already trained operators but an alternative efficient new arrangement may be selected. Key tops may be dually or singly designated.

The purpose of the article is to describe the Maltron keyboard. In doing so, Hobday makes a number of interesting claims, such as: "Sometimes the onset of RSI from keyboards is rapid (complete disability within a week), but more often it builds up slowly over months or even years."

"The flat typewriter style keyboard has been around for so long that a severe cultural shock occurs when an entirely new shape is presented as an office machine keyboard. This is especially so for trained operators who find the different physical layout of key positions to be confusing. Some gentle encouragement may be needed to help to overcome initial reluctance and to practice sufficiently to develop new reflexes. The problem occurs because the sense of key position derived from practice on the flat keyboard is no longer correct. This is a passing phase. It has been demonstrated that in as little as 25 minutes, normal speed can be attained. While this may be exceptional, it can certainly be achieved in a day or so. To help with further improvement, an adaptation training course is supplied with the keyboard. Results show that a 20% improvement in speed is easily accomplished, with a substantial reduction in errors and fatigue as expected."

Finally, the author concludes, "the fact that the design requires only a few hours of adaptation practice by already trained operators to achieve much more comfortable working conditions, and at the same time offers a significant improvement in productivity, should appeal to both operators and employers.

Hobday (1988) reported that the Maltron keyboard, which addresses the issue of ulnar deviation and pronation through splitting and lateral angling of the key fields, showed a 20% improvement in performance and a substantial reduction in error and fatigue.


Honan, Serina, Tal & Rempel, 1995

Honan et al. (1995) performed a study to evaluate the effect of keyboard design and height on continuously recorded wrist and forearm postures. Keyboard configurations were:

Keyboard A (13.5 degree opening, 5.5 degree slope and 8.5 degree tenting)

Keyboard B (12.0 degree opening, -2.6 degree slope and 10.0 degree tenting)

Keyboard C (0.0 degree opening, 8.0 degree slope and 0.0 degree tenting)

Four electronic goniometers were mounted to the subject’s arms and hands to continuously record wrist extension/flexion, ulnar/radial deviation, and pronation/supination. After all keyboard configurations were tested, subject completed a psychophysical questionnaire. Alternative keyboard B alters forearm and wrist postures toward neutral when compared to a standard keyboard. Subjects preferred the standard keyboard over the alternative keyboards based on psychophysical questionnaire results from their short duration of exposure.


Ilg, 1987

The design of keyboards is still characterized by that of mechanical typewriters. This paper presents a summary of a research project dealing with the ergonomic improvement of keyboards, carried out at the IAO in Stuttgart during the past five years. Extensive laboratory evaluations of experimental keyboards, where different design parameters were tested under real life conditions, have produced a relative optimum regarding ergonomic keyboard design. An accompanying investigations of user acceptance evaluated all realized parameters. In co-operation with a keyboard manufacturer, the results were used to design a marketable product, which may be seen as an important contribution to ergonomic keyboard design.

The article describes the development of the Marquardt keyboard. The author states that: "The main problem for the ergonomic design of keyboards is the unnatural position of the hands while typing. It has been recognized that the operation of conventional keyboards leads to tenosynovitis because of the intensive abduction, together with the static carriage of the hands, involved in the typing process. The objective of an ergonomic keyboard design is to create a human-oriented tool. The stress and strain of typing will be optimized by mechanical and informative adjustment. Increased output is not the primary objective, but will inevitably occur due to the improved ergonomics. The main criteria for success are the reduction of stress components; the improvement of output; and the improvement of operator acceptance.

During the development of the keyboard, the investigators studied 16 keyboard design parameters, using a sequential test plan. The parameters were:

  • Distance between hands
  • Geometric arrangement of keys
  • Vertical key spacing
  • Horizontal key spacing
  • Lateral inclination
  • Frontal inclination (tilt)
  • Curvature longitudinal to the fingers
  • Curvature transverse to the fingers
  • Depth of depression (coarse test)
  • Depth of depression (precise test)
  • Force (coarse test)
  • Force (precise test)
  • Key design (side-view)
  • Key design (top-view)
  • Key measurements (maximum diameter)
  • Keyboard comparison (standard vs ergonomic)

The dependent measures for the studies were keystrokes per unit time, number of errors, and user acceptance. Keyboard evaluations used 30 participants (highly proficient, medium proficient, and non-proficient typists) performing four typing tasks per parameter level. The investigators combined the dependent measures using a nonparametric procedure, the used Friedman tests to determine which level of a parameter was best. The best level for a parameter went into the next test. Some of the results of the parameter testing were:

  • Use 19mm vertical and horizontal key spacing.
  • use 8-degree lateral inclination.
  • Use 10-degree frontal inclination (tilt).
  • No need for keyboard curvature.
  • Use 4 mm for depth of key depression.
  • Use a force between 20 and 70 cN.
  • Use a keytop with minimum 30-degree concave surface.
  • Keytops should be square, with 14mm on a side.
  • All subjects found the hand-rest to be desirable.

"A further and final investigation on user acceptance with a control group which did not participate in the laboratory experiment, showed clearly that the change of keyboard design, after some practice time, will have considerable advantages for data input."


Jahns, Litewka, Lunde, Farrand, & Hargreaves, 1991

This poster reports a pilot study in which eight healthy participants and one injured individual used the Kinesis split keyboard for about eight hours over three experimental sessions. The participants completed typing exercises and word-processing sessions. Before using the experimental keyboard, the participants completed a ten-minute typing test with a standard keyboard. Because the injured sample size included only one individual, this summary will focus on the results for the healthy sample.

Performance in the initial traditional keyboard test ranged from 35 to 93 words-per-minute, with accuracy ranging from 88 to 97%. Typing speed on the split keyboard generally began to plateau after 30 minutes of fingering exercises, and averaged 90% of baseline speed within 2.5 hours. In the final test, average speed for the healthy group was 95% of the baseline. Six of the eight participants stated that they preferred the experimental keyboard overall.


Jancsurak, 1993

This is a one page review of the Marquardt ergonomic keyboard. The author states: "The keyboard resulted from extensive research at the Fraunhofer Institute for Industrial Science and Organization, Stuttgart, Germany . Investigations by the Technical University in Darmstadt, Germany, have shown that typing performance with MiniErgo is equal to or greater than that with standard keyboards. Say Marquardt’s Groff: ‘Research indicates that a high level of performance can be achieved in a relatively short time and with less effort than with a conventional keyboard.’ To arrive at the final design, more than 105 keyboard designs were evaluated using approximately 5,000 individuals (500 groups of 10) with typing skills ranging from limited to advanced. Slow-motion video of hand movement were studied. Results from electromyographic test, using sensors to study muscle activity, were carefully evaluated."


Klockenberg, 1926

This seems to be the source of the concept of a split keyboard. According to Kroemer (1972, p.51), Klockenberg "described how the keyboard layout (which has not changed much since) required the typist to assume postures of the trunk, heat, shoulders, arms, and hands that are ‘unnatural, uncomfortable and fatiguing.’ Based on detailed physiological considerations, and on experiments including time and motion studies, Klockenberg suggested a number of improvements." One suggestion was to split, separate and incline the keyboard sections for each hand to reduce muscular tension in the typist’s shoulders and arms.


Kroemer, 1972

Studies on split keyboards were carried out by Kroemer (1972) who proposed splitting the keyboard into two parts and arranging them in such a way that the hands could be kept in a more natural position. Kroemer’s halved keyboards had an opening angle of 30-degrees and a lateral declination adjustable between 0 and 90-degrees. The opening angle is defined by two lines running through the inner board of the keys Y-H-N and T-G-B. Experiments disclosed that the experimental keyboard generated less painful fatigue than traditional typewriters. (Grandjean, 1988)

Support for lateral inclination was reported by Kroemer (1972) where subjects preferred a mid-range (40 degree) forearm pronation angle. He found no significant differences existing for either typing speed or errors when using different spatial configurations of an adjustable keyboard.

In Kroemer’s (1972) first experiment, he had participants hold rods in their hands and rotate their forearms until they were at their most comfortable position. This was performed with their arms hanging vertically or laterally elevated by 15, 45 or 90 degrees. Most participants found a forearm pronation angle of about 40 degrees preferable. In a second experiment he used an experimental split keyboard with lateral declinations of 0, 30, 60, and 90 degrees with an opening angle (split) of 30 degrees. No significant differences existed for either typing speed or errors when using the different spatial configurations. The only marked difference on the preference data was that 17 of 25 participants preferred the 60-degree declination to horizontal. His third experiment compared the split keyboard, with a 50 degree split and 45 degree declination angle, to a standard keyboard. There were no significant differences between the tapping rate or heart rate. There was, however, a 7.7% error rate at the K-keyboard versus a 12.7 error rate at the standard keyboard. The participants expressed reasons for ending the session appeared to favor the K-keyboard due to less comments related to aches and pains as compared to the standard keyboard.

Abstract: (English description of Kroemer (1964, 1965a, 1965b): The standard typewriter keyboard serves as a model for keyboards of teletypewriters, desk calculators, consoles, computer keysets, cash registers, etc. This man-machine interface should be designed to allow high-frequency, error-free operation with the least possible strain on the operator. This paper discusses several feasible biomechanical improvements of the keyboard. Some experimental findings are described which support the following design concepts: (1) the keys should be arranged in a :hand-configured" grouping to simplify the motion patterns of the fingers; (2) the keyboard sections allotted to each hand should be physically separated to facilitate the positioning of the fingers; and (3) the keyboard sections allotted to each hand should be declined laterally to reduce postural muscular strain of the operator.

The paper reports three experiments:

Experiment 1: Preferred Posture of Forearms and Hands

The study focused on the preferred posture of forearms and hands. Thirty-eight female participants held a 26-gram 20-cm rod in each hand. Then with upper arms hanging vertically or laterally elevated by 15, 45, or 90 degrees, they rotated their forearms until they were at their most comfortable position. The dependent measure was the forearm angle at this position, with a handshake position as 0 degrees and a horizontal hand (relative to the horizon) as 90 degrees. With the upper arm hanging, most participants preferred a forearm pronation angle of about 40 degrees. The only upper arm elevation at which a 90-degree forearm pronation angle was comfortable was 90 degrees (straight out from the shoulder).

Experiment 2: Different Spatial Organizations of the K-Keyboard

Kroemer developed an experimental split keyboard using data from Klockenberg (1926), which he named the K-keyboard in honor of Klockenberg. He assigned letters to keys so typists typing a target sentence would use their fingers with the same frequency and motion as typing standard text with a standard typewriter. Although this was a somewhat artificial task, Kroemer stated, "The motion pattern seemed to be sufficiently complicated and general enough to represent any type of high-skill keyboard operation." A board (which has a picture of the keyboard layout) hid the keyboard from the typist’s view. The independent variable was the extent of lateral declination of the keyboard (0, 30, 60, 90 degrees) with the angle of the split held constant at 30 degrees. After completing all sessions, participants indicated which angle they preferred. Analyses of variance showed no significant differences among the experimental conditions for either typing speed or errors. On the basis of preference data, Kroemer stated, "Among the preferences voiced by the subjects, the only marked one was the 60-degree declination of the keyboards over the horizontal arrangement." 17 of 25 participants preferred the 60-degree declination to horizontal.

Experiment 3: Standard Keyboard versus K-Keyboard at 45 Degrees.

The method is almost identical to that described above. The purpose of this study was to compare the split K-keyboard with a standard keyboard (tilted 15 degrees, but with no declination). The K-keyboard in this study had the vertex of the centerpiece separating the keyboard sections was 50 degrees (rather than 30 above), and the angle of declination was always 45 degrees. Over experimental sessions there was no significant difference between the keyboards for tapping rate or heart rate. "Clearly different, however, were the number of ‘errors’ committed when using either keyboard: for every 1,000 strokes, 77 errors occurred at the K-keyboard, versus 127 at the standard keyboard". On the basis of participants’ expressed reasons for ending the experimental sessions, Kroemer concluded, "Physiological symptoms (‘aches and pains’) are given more often after work at the standard keyboard while the less specific ‘can no longer concentrate’ appears most often in connection with the K-keyboard.


Lopez, 1993

Thirty-six female touch typists participated in this study. Twenty of the typists had carpal tunnel syndrome (CTS). Typists in this group met at least two of three diagnostic criteria for CTS: a positive Phalen’s test, vibrometry score of greater than 1.5 vibration units, and a nerve conduction velocity of greater than 4.5 ms. The remaining sixteen typists were the normal control group. Participants used four keyboard designs over a two-day period: a standard IBM keyboard, a standard keyboard with a contoured wrist rest (Wave), a horizontally-split keyboard (Kinesis), and a vertically-oriented keyboard (Comfort). On the first day, the participants became familiar with the Comfort and Kinesis keyboards, performed two three-minute typing tests with the IBM keyboard to establish a performance baseline, then practiced with the Comfort and Kinesis keyboards for fifty minutes each (with order of presentation randomized). To assess accommodation (learning), the participants took three-minute typing tests at the 30-, 40-, and 50-minute points within each practice session. On the second day, the participants performed four different typing tasks with the four keyboards. The experimental design used a Greco-Latin square to balance the effects of keyboard, presentation order, and typing task. After each typing session, the experimenter made various physiologic measurements and participants completed various questionnaires. The adjustment angle of the Comfort keyboard was 60 degrees. The workstation arrangement and seating were consistent with ANSI/HFS Standard No. 100-1988. The dependent variables fell into five sets:

  • Nerve conduction velocity, vibrometry, and hand strength
  • Digit temperature and hand volume
  • Posture and discomfort
  • Accommodation and performance
  • Subjective user evaluation

Differences in nerve conduction velocity, vibrometry, and strength between keyboards and groups. Although there were statistically significant differences between the CTS and control groups, the type of keyboard had no effect on any of these physiologic measures. "The lack of statistical significance in the physiologic differences among the keyboards is important. The findings indicate the postural redesign alone does not affect physiologic functioning. It appears that an interaction of factors such as the mechanical and postural features of the keyboard as well as the workstation arrangement, job design, and physical status of the individual contribute to the development of CTS."

Analysis of the effect of keyboard design on temperature and volume change. The group differences and keyboard differences were not significant for the measurement of digit tip temperature, although digit tip temperature significantly fell as a function of time spent typing. For hand and distal forearm volume change, the groups were significantly different, with higher volumes occurring in the CTS group. The volume change differences among the keyboards were not statistically significant. Lopez (1993) concluded that the changes in digit tip temperature over time suggests that keyboard design should consider the force-displacement characteristics of the individual key. The nonsignificant differences among the keyboards suggest that keyboard geometry does not have an effect on these measures.

Posture and discomfort with different keyboard designs. To assess posture, the experimenter used a video recorder to take pictures of the participants from fixed side and back locations as they typed. The dependent measures were the joint angles of the shoulder flexion and abduction, elbow flexion, wrist flexion and ulnar deviation. For the CTS group, the type of keyboard significantly affected right shoulder abduction, right and left elbow flexion, right and left wrist extension, and right and left ulnar deviation. For the control group, the type of keyboard significantly affected right and left wrist extension, and right and left ulnar deviation.

CTS group showed significant right shoulder abduction between the IBM and Comfort Keyboards, with greater abduction with the Comfort keyboard.

Right elbow flexion was significantly different between Wave and Comfort keyboards, with the greater flexion occurring with the Comfort keyboard.

For right and left wrist extension, the Comfort keyboard had significantly greater wrist extension than the other three keyboards, which were not significantly different from one another.

Greater left and right ulnar deviation occurred with the IBM and Wave keyboards. The Kinesis keyboard had less left ulnar deviation than the Comfort keyboard. This was also true with the control group.

For the normal control group, the greater wrist extension occurred with the Comfort keyboard, and less extension occurred with the Kinesis than the IBM keyboard.

There were statistically significant differences in ratings for upper arm, forearm, and wrist discomfort. For both the CTS and normal groups, the Comfort keyboard was the least comfortable keyboard for all three areas (upper arm, forearm, and wrist).

The CTS group found the Kinesis keyboard to be the most comfortable for all three area. The IBM received the most favorable ratings from the normal control group for minimizing upper arm discomfort. The normal control group rated the Kinesis most favorably for minimizing forearm and wrist discomfort.

Typing speed on the Comfort keyboard was higher than the Kinesis; however, it never reached the IBM or Wave level of speed -- Note: The speed difference did not reach a level of statistical significance.


Loricchio, 1992

The IBM Design Center in Boca Raton studied the operating-point key force for a portable computer keyboard. Alden, Daniels, and Kanarick (1972) reported that typists prefer operating-point key forces of between 25 and 150 grams. We compared different key forces that fell within the range recommended by Alden et al. The only difference between the keyboards we studied was the amount of force required to activate the keys. The first keyboard (58 keyboard) required 58 grams of force to activate the keys. The second keyboard (74 keyboard) required 74 grams of force to activate the keys. Sixteen skilled typists used both keyboards to enter text. Input speed was significantly faster on the 58-gram keyboard. A significant number of typists preferred the 58-gram keyboard. The results suggest that the optimal key force for portable computer keyboards is less than 74 grams.


Lu, 1983

The theme of this magazine article was "Drive for office productivity overlooks keyboards.": Part of the article addressed the Maltron keyboard: Lillian Malt rethought not only the key assignments but he shape as well and came up with the PCD-Maltron keyboard in 1976. Its sculpted contours conform to the finger positions of a relaxed hand. The letter assignments are based on a more thorough analysis of the English language than Dvorak’s. Malt also used different ergonomic criteria, believing, for example, successive keying with different fingers of the same hand to be faster than using alternating hands. This, and other ergonomic considerations, are not well-established. Unfortunately, the BACKSPACE and TAB keys don’t yet have a final resting place. . . Barely 70 Maltron keyboards have been sold in five years, most of them for evaluation purposes. (p.23).


Marek, Noworol, Wos, Karwowski, & Hamiga, 1992

Sixteen healthy females with limited typing experience participated in the study. The study included two types of keyboards: a standard keyboard and one constructed according to the experimental data of Kroemer (1972), Nakaseko et al. (1985), Zipp et al. (1983), and Hamiga (1989). Each participant typed the same test with both keyboards for 15 minutes, with half of the typists using the standard keyboard first. Participants rested for at least two hours between sessions. The workstation did not include wrist or forearm supports, although the experimental keyboard had a large wrist support.

The dependent measures were electromyographic (EMG) recordings from the left and right trapezius (M. trapezius pars descendeus) and extensor (M. extensor carpi radialis brevis et longus) muscles, and participants’ perceived levels of muscular tension in the neck-shoulder area and forearms. For these subjective ratings, participants marked a position on a 10-cm long, unmarked scale with end-points of very relaxed and very tense.

The EMG measurements indicated a significant difference (t-test with p<.05) in muscle loading in favor of the split keyboard for the trapezius (shoulders-back), but not for the extensor (forearm). The same pattern appeared in the subjective ratings.

The significant effect found in this study, muscle loading and reported discomfort in the shoulders, did not appear in the other experiments that measured this (Kroemer, 1972; Nakaseko et al., 1985; Brigham and Clark, 1986; Thompson et al. 1990). These same studies found differences in the arm-wrist comfort, while Marek et al. study found no differences.


Morelli, Johnson, Reddell & Lau, 1995

The proliferation of varying computer keyboard designs may pose problems to those who specify, purchase and ultimately use such devices. Are any of the "best" for my work? Will actual users derive any benefit from the? To assist in addressing such issues, we explored an approach to determining if any of three "alternative" keyboards provided a benefit to employees by increasing user comfort. A total of 34 employees participated in this study, each using an "alternative" keyboard for one week at a time while performing their actual work. After using each keyboard, including their standard "101" keyboard, subjects completed a questionnaire of seven psychophysical measures relating to the comfort and use of the keyboard. Questionnaire responses were tabulated and a Repeated Measures Analysis of Variance conducted. Statistically significant differences were found among and between the four keyboards on four of the seven psychophysical measures. Overall results showed little differences in user assessments between three of the keyboards, with the fourth keyboard assessed less favorably when significant differences were found. The results suggest that user assessments can produce significant results between keyboards, and that on one "best" keyboard exists for any given set of tasks and group of users.

Results linked to the two measures of force requirements yielded identical results, i.e. no differences among the four keyboards. This would suggest that the activation forces for these keyboards appears to be similar based on the subjects’ responses. Whether key activation forces have been designed specified optimally cannot be determined here.

The perceived reaction to the mechanical characteristics showed similar results across and among the four keyboards, with keyboard "C" generally assessed as less desirable than the others. Comments supplied by the users in the follow-up interviews suggested that the awkwardness of reaches and the small size of function keys on this keyboard probably contributed to much of this difference.

The statistical differences in the assessments of postural demands, again singled out keyboard "C" as rated unfavorably in the perceptions of the users. However, the written comments and subject follow-up interviews strongly suggested this difference did not primarily involve the hand and wrists, but reflected whole body discomfort. Multiple comments (15%) made reference to needing to peer down into the key wells on keyboard "C", which induced undesirable upper body postures. Almost 50% of responses commented on the awkwardness of toggling back and forth between numeric and alpha entry via a function key.

The lack of differences between the keyboards regarding "aches and pains" is probably attributable to both the relatively short time span of the tests, although these tests were much longer than virtually all other tests cited in the literature, and the care taken by the ergonomists in set-up and adjustment of each work place for each individual participant.

The generally favorable assessment of the standard keyboard may be attributed as much to its familiarity as to its basic design, which was suggested in the follow-up interviews.

A confounding element in the results was the fact that adjustable articulating forearm supports were only provided with keyboard "A". This effect could not be accounted for or controlled in the results. The follow-up interviews did suggest the importance of these arm supports, with approximately 60% of participants indicating they would like to have the arm supports with their standard keyboard or any of the other keyboard they most preferred.

The results confirm that psychophysical assessments can yield significant differences among keyboard ratings when used in performing "real" work, and suggest that users, as a group, will not identify one "best" keyboard for performing a given task. These findings should not be construed to indicate that any of the keyboards assessed by this group of subjects is "better" or "worse" than any other, except for the specific tasks involved in this study.

The follow-up interviews strongly suggested that the subjects had high preference for a modular numeric keypad, which influenced their assessments of keyboards "A" and "B". Had a modular numeric keypad been used with keyboard "C", the assessments could have been different.

The known and expected variances in any worker population and the results found here would indicate that user assessments can play a role in selecting keyboards. In addition, a mix of keyboards selected by user assessments, possibly along with adjustable articulating forearm supports and care to adjust work station components, is an approach to attaining high user acceptance and comfort.


Morita, 1989

This article describes the ergonomic philosophy and development of the Bungo EMII Japanese Word Processor Keyboard. The ergonomic philosophy is essentially the same as that published by Kroemer (1972) and Nakaseko et at. (1985), although the author does not cite these works. The article does not discuss any experiments with the Bungo split keyboard.


McAlindon (1994)

In an experiment using limited character sets, there was no significant difference detected between the learning rates of the Keybowl and keyboard subjects. In fact, the rates for the Keybowl typists in sessions 15 and 16 were almost identical to session 1 and 2 rates of keyboard typing. This suggests that experienced typists could reach comparable performance on the Keybowl as compared to the keyboard in a relatively short period of time. Ergonomic evaluation of the Keybowl and keyboard was performed by measuring the left and right hand wrist deviations using electrogoniometers. There was a significant difference detected in wrist movements between the two subject groups for both flexion/extension and ulnar/radial movements of the left and right hand. Overall wrist variance was reduced by an average of 81.5% in the flexion/extension plane and 48% in the ulnar/radial place. The Keybowl user’ wrist deviations averaged 3 degrees in the flexion/extension plane and 3 degrees in the ulnar/radial deviation plane. The keyboard user’ wrist deviations averaged 15 degrees in the flexion/extension plane and 6 degrees in the ulnar/radial deviation plane. These preliminary results indicate the Keybowl typists may be able to quick achieve performance levels comparable to that of QWERTY typists, while eliminating finger movements and drastically reducing wrist movements.


Nakaseko, Grandjean, Hunting, & Gierer, 1985

Experiments with split keyboards: Following this line of research, Grandjean et al. (1981), Hunting et al. (1982) and Nakaseko et al (1985) developed an adjustable model of a split keyboard and studied the preferred settings of opening angles, lateral sloping and distances between the split keyboards on 51 subjects. Typing with the split keyboard with preferred settings decreased the lateral adduction of the hands shown in Figure 68, reduced discomfort and increased feelings of being relaxed in the neck/shoulder/arm/hand area. (Grandjean, 1988)

Effects of a large forearm/wrist support: The use of a large forearm/wrist support was associated with a declined sitting posture of the subjects and with an increased pressure load of the forearm/wrist on the support, reaching mean values of nearly 40 N (4 kg). Such weight transfers onto the support will strongly decrease the load on the intervertebral discs. (Grandjean, 1988)

Preferred settings of split keyboards: Of 51 subjects, 40 preferred the split keyboard with the following characteristics (Grandjean, 1988):

  • A 20-degree angle between the two half-keyboards.
  • A distance of 95 mm between the two half-keyboards (measured as distance between the keys G and H).
  • A 10-degree lateral sloping of both half-keyboards
  • A hand-configured design of the keys

A prototype of such a keyboard is shown in Figure 69. A commercial model was presented at the Ergodesign 84 conference by Buesen (1984). (Grandjean, 1988)

A keyboard concept based on biomechanical considerations was studied with 51 trained typists. The keyboard is split into two half-keyboards. An adjustable model allowed study of the preferred settings of opening angles, lateral inclinations, and distances of the split keyboard. The preferred split keyboards decrease the lateral deviation of the hands, and the use of a large forearm-wrist support is associated with a backwards leaning of the subjects and with an increased pressure of forearm-wrist onto the support. After the typing tasks, about two-thirds of the subject asserted that they preferred the split keyboard models. Less pain and an increased feeling of relaxation were reported by the subjects when operating the split keyboards.

The article reports the results of three studies: a preliminary field study, a preliminary study of the acceptance of a new keyboard concept, and a study of the effects of their new keyboard design.

The preliminary field study focused on the reported ‘daily pains’ and medical examinations of three groups of employees who used keyboards )data-entry terminals, conversational terminals, and full-time typists) relative to a control group of traditional office workers. The keyboard users had a greater incidence of pains (both self-reported and from the medical examination) than the traditional office workers. Kakasek et al. point out: "The incidence of complaints or medical findings was increased at workstations with keyboard levels that were low in relation to the floor and with keyboard levels that were high above the desk. Complaints were also greater at workstations that provided no support for forearms or wrists and at those in which workers assumed ulnar deviations of the hands exceeding 15 to 20 degrees." (p. 178).

Preliminary study of the acceptance of a new keyboard concept. Participants in this study (20 trained female typists) used four or five keyboards for 45 minutes each to complete typing tasks. Two were traditional keyboards, one with a forearm-wrist support and one without. Three experimental keyboards were split, all with forearm-wrist supports. The experimenters varied the opening angle of the split (0, 25, and 35 degrees), the distance between the G and H keys (6.5, 9.5, 11.5 cm), and the lateral inclination of the keyboard halves (0, 10. and 10 degrees). The frontal inclination of all three experimental keyboards was 10 degrees. Ten participants used all keyboards, and the other ten participants used all the keyboards except the traditional keyboard without forearm-wrist supports. The authors did not specify the order of presentation of the keyboards. The authors concluded, " the results show that subjects gave a clear preference to the split keyboard with a lateral inclination of 10 degrees and an opening angle of 25 degrees." (p.179)

Effects of the new keyboard design. In the final experiment, 31 participants (1 male and 30 female trained typists) typed for 30 minutes with each of three keyboards. One keyboard was a traditional keyboard with a large forearm-wrist support, one was an experimental split keyboard with a large forearm-wrist support (the keyboard from the previous experiment, with 10 degrees lateral inclination and 25 degrees opening angle), and one was the same experiment keyboard with a small forearm-wrist support. Participants used the keyboards in random orders. After using each keyboard, participants completed a questionnaire to indicate whenever they felt pain in different parts of the upper limbs and shoulders. They also reported their feelings after each trial, using a seven-point scale with the end points of very relaxed and very tense. The experimenters also recorded the amount of pressure the participants placed against the forearm-wrist supports and made a number of body posture measurements. Finally, the participants ranked the keyboards according to their preference.

There were no statistically significant differences among the keyboards or reported pains in the neck-shoulder and arm-hand areas. For the relaxed-tense ratings, the keyboards did differ for the arms and hands and the back (X2 test, p<.05), with the experimental keyboard with the large forearm-wrist support having the most favorable ratings. However, none of the average ratings exceeded the center point of the seven-point scale (Neither-Nor). Participants always place less pressure on the small forearm-wrist support for the duration of a 30-minute session. During the first 15 minutes of a session, participants places less pressure on the large forearm-wrist support when using the traditional keyboard, but for the next 15 minutes the pressure was equal for the two keyboards with large forearm-wrist supports. The authors state: "the traditional keyboard was preferred by a scarce 30% whereas more than two-thirds prefer one of the two experimental keyboards" (p. 185). The authors did not perform an overall statistical test on the data, nor did they report any performance data.


Noyes, 1983

The main topic of this paper was a review of alternative key layouts. Regarding split keyboards, the author provided a cursory review of Kroemer (1972).


Potosnak, 1990

The Handbook of Human-Computer Interaction is an influential reference book. In the chapter on keys and keyboards, Potosnak includes a brief review of the split-keyboard literature under the section "Innovations in Keyboard Design." Potosnak accepts the work of Kroemer (1972), but provides a more critical review of Nakaseko et al. (1985): "Typists found the split keyboard acceptable and favored the design with lateral tilt of 10 degrees and an opening angle of 25 degrees. Although performance data have not been reported, preferences and measures of arm and hand positions show a strong effect due to the use of a large forearm-wrist support. In fact, a traditional keyboard with a large support was preferred by more typists than the experimental keyboard with a small support. The use of a large forearm-wrist support might also result in a less stressful posture.


Rempel, Smutz, So, & Armstrong, 1994

Indirect evidence indicates that the pathophysiologic mechanism of CTD involves repeated elevation of carpal tunnel pressures (CTP). These elevated pressures may lead to ischemia and flexor synovial tissue edema which in turn leads to further elevations in CTP and eventually nerve blockage.

Carpal tunnel pressure was measured by means of a saline-filled blunt-tipped, multiperforated 20-gauge catheter inserted percutaneously into the carpal tunnel of the non-dominant hand. The catheter was attached at its proximal end to an in-line pressure transducer which was maintained at the same elevation as the carpal tunnel. Ulnar/radial deviation and flexion/extension of the wrist were monitored using a two-channel electrogoniometer secured to the dorsal surface of the hand and forearm.

The non-dominant hand and forearm were placed palm down in the test fixture. The subject was then instructed to press down on the pinch meter with the tip of the index finger with 0, 6, 9 and 12 N force.

Results indicate that with the wrist in a neutral position, increased fingertip loading results in increased CTP. Furthermore, the CTP to load relationship remained when the wrist was deviated from its neutral position to any of the four directions tested.


Rose, 1991

A substantial proportion of the problems in keyboard operator ‘Over-use Syndromes’ occur in the wrist and finger extensor muscle group. Biomechanical analysis shows these muscles to be subject to substantial sustained static (isometric) muscle contraction during the work task. This study measured the maximum relaxed finger press forces for 60 subjects in three arm support methods, in order to predict what the minimum keypress force should be to permit finger support sufficient to facilitate relaxation of finger extensor muscles. It was postulated that the minimum key activation force should accommodate the 95 percentile predicted population relaxed finger weights. The predicted force of 0.8 Newton is within limits previously proposed for performance criteria, and which have been found practical commercially.

The tendency to rest wrists on keyboard support surfaces is common in order to reduce upper arm and shoulder fatigue. The results of this study indicate that finger weights will be increased with thin keyboards, further increasing muscle tension necessary to avoid accidental key actuation unless the key actuation force is greater than finger weights. In order to accommodate the mean relaxed finger weight, at least 0.5 N key activation force would be required. To design for the 95th percentile range, at lest 0.8 N is indicated to allow the finger to relax onto the keys, so reducing the extensor muscle force to below 1% MVC without fear of accidental key activation

Should arm or wrist support be provided close to or level with the key tops, 0.7 N key activation force is indicated to provide resting finger support. for no wrist or arm support to avoid accidental key actuation, 1.1 N force would be required. A statistically significant difference in relaxed finger weights between make and female subjects was not found (P<0.05).

Static contraction levels of finger muscles when insufficient finger weight support is provided by keys are in excess of the recommended levels and must therefore be considered a significant potential contributor (when sustained) to extensor muscle fatigue and potential Over-use Syndrome outcome. This finding has significance for the design and operation of keyboard devices, particularly those requiring sustained operation - e.g., alphanumeric entry keyboards, mousse.

The high levels of isometric muscle contraction in evidence with keyboard operating postures also indicates that training and job design should encourage the resting of fingers. Tasks should be introduced which are away from keyboards at suitable intervals to further facilitate removal of isometric muscle contraction, particularly for tasks where small finger hyperextended postures are in evidence. Resting the hands every 15-20 min in their neutral posture for fatigue recovery, thereby reducing the stress on the extensor muscles, is a training strategy which has been successfully employed where key actuation forces are less than the above recommendations.

Improved split field keyboard designs would reduce static contractions for small finger extensors, forearm printing and wrist deviating muscles, provided wrist extension is eliminated by arranging the face of the keyboard appropriately in relation to the operator’s forearms.


Smith & Cronin, 1993

A comparative study was conducted to determine the differences in user muscle load, posture, performance and preferences for a new technology keyboard (the Kinesis) compared to a standard (traditional) keyboard. The study consisted of requiring 25 test subjects to key text and random letters for two hours on each keyboard. Results demonstrated that hand postures (deviation and extension) and muscle load were better on the Kinesis keyboard. Text entry throughput was greater on the traditional keyboard, although there was no significant difference in errors between the two keyboards. Subjects preferred the Kinesis keyboard for comfort and usability.

This experiment had two goals. One was to determine if the Kinesis keyboard achieved its design goal of reducing muscle load during typing, and the other was to determine if the Kinesis keyboard would meet the requirements of the International Organization of Standardization (ISO) draft ergonomic standard 9241. In this study the researchers simultaneously measured postures, muscle load (EMG), typing performance, and subjective impressions. Twenty-five participants (20 women, 5 men, all with several years typing experience) from a temporary employment agency used both a standard and a Kinesis Ergonomic Keyboard Model 100 to type normal and random text (as specified in the ISO 9241 draft). The researchers specified that the workstation chair and keyboard tray were adjustable, and that the Kinesis keyboard (but not the standard keyboard) had a built-in palm rest. Eleven of the 25 subjects participated in the EMG measurement part of the study. All participants practiced with he Kinesis for seven hours the day before using it in the comparative study. In the comparative study, half the participants used the standard keyboard first and half used the Kinesis first.

The results showed significantly lower muscle load for the Kinesis keyboard for the hand (extensor communis digitorum) and wrist (extensor carpi ulnaris), but not for the arm (pronator radii teres) or shoulder (deltoid) muscles. Participants typed significantly faster with the standard keyboard. The difference in errors between the keyboards was not significant. Participants preferred the Kinesis keyboard for comfort, but preferred the standard keyboard for performance.

The researchers concluded that the Kinesis met its design goal for improved comfort. However, because typing speed on the Kinesis keyboard was 1.5 words per minute slower than the target established by the ISO 9241 draft standard (that the experimental keyboard should not be more than .75 standard deviations slower than the average for the standard keyboard, given a appropriate sample size), the Kinesis would not meet the ISO requirements. They suggested that allowing more practice (about 40 hours) might bring user performance into the ISO acceptable range.

This paper makes a substantial contribution to the split-keyboard literature. The authors measured a wide variety of dependent variables, and, with the exception of the preference data, provided tests of statistical significance for their findings. Consistent with other research, the split keyboard seemed to require less muscle load to operate, but was substantially slower than the standard keyboard. The researchers did not force the participants to state which keyboard they preferred overall, so the preference outcome depended on whether the participants rated keyboard comfort (favoring the split keyboard) or performance (favoring the standard keyboard).


Smutz, Serina, & Rempel, 1994

Part of a special issue on ergonomics in telecommunications. A system to determine the effectiveness of alternative keyboard designs that are aimed at minimizing fingertip force and awkward wrist postures. The new system utilizes a 4-way adjustable prototype keyboard. Measurements of fingertip impact force, wrist position, productivity, and comfort and ease of use are used to evaluate the effectiveness of each keyboard configuration. Instrument keys with strain gages located between the key cap and the key switch are used to measure fingertip force. Electronic goniometers attached to the forearms and hands of the test subjects are used to measure wrist position. A computer program determines productivity, and a questionnaire based on the Borg rating scale is used to evaluate comfort and ease of use. Validation trials indicate that the system is capable of measuring fingertip force and wrist angle with accuracy and repeatability.


Sommerich (1994)

Sommerich (1994) measured CTP using a commercial keyboard in both standard and split conditions. She found that splitting the keyboard eases ulnar deviation and also results in less pressure in the wrist’s carpal tunnel.


Sommerich, 1995

Sommerich et al. (1995) performed an electromyographic investigation of finger and wrist muscle activity during typing. Examination of the data revealed substantial activity of the extrinsic extensor, a muscle which is ignored in many existing biomechanical finger models. This paper describes activity of the extensor muscle during typing, in absolute terms and relative to activity of the extrinsic flexors. Amplitude probability distribution analyses demonstrated that static extensor activity exceeded 5% MVC for all subjects. Two subjects exhibited pronounced patterns of coactivity in the extrinsic extensor and flexor muscles. Biomechanical modeling efforts demonstrated similar force contributions from the extrinsic extensors and flexors. Based on these results, neglect of finger extensor activity would result in underestimation of finger joint loading.


Szmanda, & Szmanda, 1992

This paper reports the development and design considerations of the Health Care Keyboard (HCK). Although the paper contains no research on the HCK versus standard or other split keyboards, the authors state: "Currently, the HCK is undergoing ergonomic/human factor research at the University of Wisconsin-Madison’s Department of Industrial Engineering, under the direction of M.J. Smith. The U.S. Internal Revenue Service is also planning a major evaluation in 1992 involving 40 HCKs. Additional research will be conducted to measure the impart of the invention on comfort, worker satisfaction, productivity and workers compensation claims."


Tessler, 1993

This article provides a fairly extensive review of the features of the Apple ergonomic keyboard. The author conducted an informal study with his fellow employees at Macworld, with these results: "Nine Macworld employees -- men and women representing a range of hand sizes and experiences with RSIs -- each used one [an Apple ergonomic keyboard] exclusively for a minimum of two weeks. Then they completed a detailed questionnaire comparing the Apple Adjustable Keyboard with the Extended Keyboard. We evaluated the keyboard on 27 points -- from overall appearance to key layout and activation pressure. The Macworld testers rated the Apple Adjustable Keyboard inferior to the Extended Keyboard in 14 categories, equal in 9, and better in only 4.


Thompson, Thomas, Cone, Daponte and Markison, 1990

Thompson, Thomas, Cone, Daponte and Markison (1990) also reported minimized objective EMG activity and subjective discomfort with split keyboards when the wrists were in a neutral position. This finding is important since poor wrist posture not only affects muscle activity, but also carpal tunnel pressure (CTP) (Weiss, Bloom, Gordon, & Rempel, 1992; Honan, Serina, Tal & Rempel, 1995) which may contribute to keyboard-related CTD injuries.

Thompson, et al. (1990) using EMG measurements showed that muscle activity is reduced as the forearm approaches a more neutral posture. They also noted a slight increase in typing speed in their experiments as the keyboard configuration moved farther away from the flat, standard design.

This keyboard allows the user to adjust, over a large range of angles, the amount of split (opening angle) and the angle of elevation for the two halves of the QWERTY part of the keyboard. The experimenters had eight participants (experienced female typists) type with four keyboard configurations: (1) Standard (no lateral inclination or opening angle), (2) Flat/Angle (no lateral inclination with an opening angle varying from 15 to 30 degrees), (3) 30-Degree Slope (same as Flat/Angle, but with the lateral angle set at 60 degrees from the horizontal). During the typing sessions, the experimenters measured four muscle groups (extensor carpi ulnaris, extensor digitorum communis, flexor carpi ulnaris, flexor digitorum sublimis) with surface-electrode electromyography (EMG). Typing trials consisted of a 15-minute learning period followed by a 15-minute typing test. According to the authors, the experimental protocol was:

"Each subject was given a series of medical history and occupational experience questionnaires to complete. They were then introduced to the variable geometry keyboard and asked to use it in its normal flat, linear configurations (standard keyboard shape) to establish a base rate for comparison purposes . . . Experimental runs were sequenced by keyboard geometry over subjects, using a Latin Square design, to avoid learning and fatigue bias. Thus, two of the eight subjects did each keyboard geometry first, two did each one second, and so forth."

The dependent measures for the study were average typing speed (words per minute), average muscle activity (mV), relative muscle productivity (mV per words per minute), and subjective preference (apparently based on a six-point scale, the rescaled to a range of -1 to +1). The authors concluded that the best opening angle was 18 degrees with a 30- to 60-degree lateral angle.


Zipp, Haider, Halpern, & Rohmert, 1983

Zipp, et al. (1983) described three experiments where: (1) a participant typed continuously for 60 minutes which resulted in increased muscle activity reflecting fatigue, (2) a participant moved his wrist and forearm to indicate the tolerable ranges for pronation and ulnar abduction. They concluded that the optimal range for pronation was 0 to 60 degrees, and for ulnar abduction was 0 to 15 degrees. These were outside the normal positions observed for typewriting at a workstation, which were observed at 90 degrees for pronation, 20 to 25 degrees for ulnar abduction. (3) a thin bar was fixed at various angles between two stands to simulate a keyboard’s home row. The angles of the simulated split keyboard’s lateral inclination were 10, 20, and 30 degrees, and the angles of simulated key field rotation were 13 and 26 degrees. They concluded that a split keyboard should have two symmetrical key fields, each rotated in the horizontal plane at an angle of 10 to 20 degrees and laterally inclined from 10 to 20 degrees.

Zipp et al. (1983) determined the affects of different keyboard configurations by using EMG measurements. They found that by splitting the key fields, the strain on hand, neck, shoulder and arm muscles were reduced. Zip et al. (1983) found that even a lateral angle as small as 10 degrees can significantly reduce muscle activity.

The authors briefly reviewed field studies that indicated keyboard operators (both typists and VDT users) have more musculoskeletal problems and complaints than their share in the working population. "Bearing these results in mind, one may conclude that the finger muscles are the only actual effectors in typewriting, while the body as a whole functions as a support holding the fingers in their work position" (p. 117). They then review Kroemer’s (1972) work with split keyboards, and state that the purpose of their research was to replicate partially Kroemer’s work with electromyographic (EMG) measurements because Kroemer measured operators’ productivity and subjective judgments. A photograph of a workstation (p. 117) shows a typing workstation with no auxiliary ergonomic support devices (such as arm or wrist supports).

They conducted three experiments, with one participant per experiment. They recorded the muscle action potentials with amplified signals from surface electrodes. Electrodes recorded activity from M. trapezius, M. biceps, M. flexor carpi ulnaris, M. extensor carpi ulnaris, M. pronator teres, and M. flexor carpi radialis. They interpreted increased in electrical activity as an indicator of increased muscle activity, and used correlation coefficients to quantify the increase. Coefficients significantly higher than zero indicated localized muscular fatigue.

In the first experiment, the participant typed continuously for 60 minutes. Five of six recorded muscles showed an increase in activity, reflecting fatigues in the shoulder, and arm muscles.

In the second experiment, the participant (not at a typing workstation) moved his wrist and forearm to indicate the tolerable ranges for pronation and ulnar abduction. The electrical activity increased as a function of approach to extreme positions. the optimal range for pronation was 0 to 60 degrees, and for ulnar abduction was 0 to 15 degrees outside the normal positions observed for typewriting at a workstation like that shown on page 117 in the article (90 degrees for pronation, 20 to 25 degrees for ulnar abduction).

The purpose of the third experiment was "to gain insight into the strain imposed on the upper limb muscles and the effect of an altered keyboard position" (p. 119). The authors used a thin bar fixed at various angles between two stands to simulate the home row of a keyboard. The participant’s task was to grasp the bar lightly without resting on it. The angles of the simulated split keyboard’s lateral inclination were 10, 20, and 30 degrees, and the angles of simulated key field rotation were 13 and 26 degrees. Lateral inclination of 10 degrees significantly reduced electrical activity, and rotating the key fields reduced the strain of the hand abductor muscle (M. extensor carpi ulnaris) and the neck, shoulder and arm muscles.

The authors concluded that a split keyboard should have two symmetrical key fields, each rotated in the horizontal plane at an angle of 10 to 20 degrees and laterally inclined from 10 to 20 degrees. "The work load should be the decisive factor for determining where to introduce the proposed keyboard. Evidently, the proposed keyboard is assumed to alleviate the postural constraints imposed on full-time touch-typists. The acceptability of a new keyboard may be better tested then at workplaces employing full-time typists and data-entry operators".

Prev ] Home ] Up ] Next ]

Return to Alternative Keyboard FAQ



Last Updated: 10/05/99