Printer Friendly

Effects of a split keyboard design and wrist rest on performance, posture, and comfort.


For more than 20 years there has been concern that working at computer terminals could lead to musculoskeletal discomfort and disorders (Bammer & Martin, 1988; Bergqvist, Wolgast, Nilsson, & Voss, 1995; Bernard, Sauter, Peterson, & Hales, 1992; Cakir, Hart, & Stewart, 1979; Gunnarsson & Ostberg, 1977; Hales, Sauter, Peterson, & Bernard, 1992; Hunting, Laubli, & Grandjean, 1981; Smith, 1984; Smith, Stammerjohn, Cohen, & Happ, 1981). A variety of anatomical sites have been isolated for particular musculoskeletal discomfort and disorders, including the neckshoulder region (Hunting et al., 1981; Hagberg, 1981; Ohlsson, Attewell, Johnsson, Ahlm, & Skerfving, 1994), the back (Bongers, de Winter, Kompier, & Hildebrant, 1993; Grandjean, 1979; Mandal, 1982), the arm (Bernard et al., 1992; Hales et al., 1992), and the hand and wrist (Bernard et al., 1992; Hales et al., 1992). Increased reports of repetitive strain injury (RSI) in Australia (Bammer & Martin, 1988, 1992) and the United States (LeGrande, 1992; LeGrande & Eisen, 1988) have focused interest on keyboard users.

Several keyboard designs take into account the posture of the shoulders, arms, wrists, and hands when keying. Some ergonomists believe that the conventional flat keyboards, which require the hands to be held with the forearms in constant pronation (palms down), may be a source of biomechanical strain (Kroemer, 1972; Nakaseko, Grandjean, Hunting, & Gierer, 1985). To compensate for this pronation, operators tend to abduct their shoulders, which moves the elbows away from the body (Rose, 1991). This is in contrast to what designers have hypothesized to be the "natural" posture of the arms, wrists, and hands when they are allowed to hang free at the sides. Because one cannot use a conventional keyboard with the hands in this natural posture, a laterally sloping, split keyboard has been suggested (Kroemer, 1972; Nakaseko et al., 1985; Rose, 1991). It is believed that a split keyboard will allow the hands, wrists, and arms to be positioned in a more natural posture to reduce pronation.

It has been proposed that while typing, wrists held in extreme extension or flexion (Szabo & Chidgey, 1989) or radial or ulnar deviation (Thompson, Thomas, Cone, Daponte, & Markison, 1990) exert increased pressure on the nerves, blood vessels, and tendons passing through the carpal tunnel as well as on the extensor muscles. Rempel and Horie (1994) showed that as the wrist's angle of extension increased, carpal tunnel pressure increased. Some believe that this increased pressure can cause pathology by diminishing microvascular flow to the tendons and median nerve, increasing capillary fluid leakage, and causing edema (Rempel & Horie, 1994). In theory, if the wrist is held in a more natural posture, this pressure should be minimized. Sommerich (1994) showed that reducing the ulnar deviation of the wrist by means of using a split keyboard does in fact reduce carpal tunnel pressure.

Such a keyboard may be split in half, allowing each half to be positioned independently so that the wrists are not pronated and the shoulders not abducted. However, care must be taken when determining the degree of slope because it has been demonstrated that a high degree of slope in a flat keyboard may force the typist to increase wrist extension to reach the home row (Powers, Hedge, & Martin, 1992). Therefore, it is possible that the sloping split keyboards should be placed on lower work surfaces than flat keyboards in order to avoid wrist extension and shoulder abduction.

One split keyboard designed to address wrist posture factors is the TONY![TM] keyboard (Thompson et al., 1990). This keyboard retains the QWERTY layout with a laterally sloped, split keyboard design and a separate numeric key pad. Optimal angles for wrist posture, to reduce arm muscle tension, have been proposed from tests of this keyboard.

Another split keyboard designed to provide flexibility in wrist postures is the Comfort Keyboard[TM, which also uses the standard QWERTY layout, with the keyboard split into three sections. The sections can be moved in several dimensions, periscope in and out, twist and turn, and be set to any slope angle. The intent of the design is to allow users to adjust the keyboard to any configuration that is comfortable. This makes it possible for every user to adjust the keyboard so that the arms and wrists are in neutral postures. The concepts behind this design are that the keyboard will improve postures when adjusted for each individual and that people will be able to learn to type effectively in these new postures so that performance will not suffer. Several studies have reported postural benefits, such as reduced ulnar deviation and wrist extension (Burastero, Tittiranonda, Chen, Shih, & Rempel, 1994; Cakir, 1994; Sommerich, 1994), as well as the comfort benefits of reduced aches, pains, and tiredness (Chen et al., 1994; Marek, Noworol, Wos, Karwowski, & Hamiga, 1992) when using a keyboard with a split design.

Another way to try to increase comfort and posture while typing is to use a wrist rest and/or forearm support. By resting the palms or wrist on a wrist rest, the shoulders are relieved from having to support them. This could lead to reduced shoulder discomfort while keying. In addition, using a wrist rest could reduce wrist extension while keying as compared with resting one's wrists on the edge of the work surface. Powers et al. (1992) had typists use both a negative-sloping keyboard support with a broad wrist rest and a full-motion forearm support with a traditional keyboard. No postural differences were found between those using only a traditional keyboard and those using the traditional keyboard in combination with the full-motion forearm support. However, those using the negative-sloping keyboard support experienced decreased wrist extension compared with those using the traditional keyboard only. There were no differences between conditions with respect to radial or ulnar deviation.

We conducted the current study to investigate whether the Comfort Keyboard - which will be referred to in this paper as the split, adjustable (SA) keyboard - can allow users to maintain performance and increase musculoskeletal comfort compared with a traditional flat keyboard. If participants can adapt to the changes in postures required by this SA keyboard and perform at levels equivalent to their performance on conventional keyboards, the new design may be feasible for use at the workplace. If the changes in posture required by the new keyboard lead to problems with performance or decrease comfort or usability, the new design, however theoretically correct, may have limited utility.

Recently, Swanson, Galinsky, Cole, Pan, and Sauter (1997) conducted a similar study using clerical workers and five keyboard conditions. In their study workers' performance did drop initially with two of the alternative keyboards. However, by the second day of use, performance increased to levels similar to those seen with the standard flat keyboard. In addition, there were no differences between keyboard conditions for musculoskeletal discomfort or fatigue; differences did occur over the workday, however.

In the current study, the SA keyboard was compared with a conventional, flat keyboard in terms of performance, posture, and comfort. We hypothesized that the type of keyboard and the use of an adjustable wrist rest would influence typing performance, accuracy, and musculoskeletal posture and pain.



Over a four-month period, we recruited 24 professional touch typists who were employed by temporary help agencies for this study. These typists were paid their normal wage for participation in the study. All were required to type at least 55 words/min and have five or fewer typing errors in a 5-min typing test. The participants were screened for any history of arm, hand, wrist, shoulder, neck, or back cumulative trauma disorders. Screening was conducted by asking the typists' employers whether the typists had any medical history of these conditions and by asking the typists about their medical history to confirm the reports of their employers. A typist with any of the aforementioned problems would not have been allowed to participate, but none of our participants had to be rejected on the basis of medical history.

Four participants were dropped because of their inability to meet the time schedule of the study, another was dropped after failing the typing test, and one other was dropped because of hardware problems during one day of testing. Of the remaining 18 participants, 3 were men and 15 were women. They ranged in age from 18 to 49 years old, with an average age of 33.5 years. Typing experience ranged from 6 to 32 years, with the average being 15.9 years.

Experimental Design

The experiment was conducted over five consecutive days, 5 h each day. On Day 1 each participant was given 4 h of training to become familiar with each keyboard, the experimental procedure, and the task. Each participant then performed a text entry task for three more days, in four 1-h typing sessions each day with 10-min breaks between sessions. Participants completed questionnaires at the start and end of each day. On the fifth day they were allowed to configure and type on the SA keyboard according to their preference.

The experiment was designed to compare performance and comfort on the SA keyboard with that on a standard flat keyboard. Average key actuation forces (M = 0.56 N for flat and M = 0.57 N for SA) and characteristics between keyboards were approximately equivalent, and general key layouts were identical. The effect of the presence of a wrist rest on user comfort and performance was also assessed. The experiment was a 3 (keyboard day) x 2 (wrist rest use) x 5 (sequence) x 4 (hour) mixed design. The between-subjects independent variables were wrist rest use and sequence.

There were three possible sequences of testing over the three days: (a) adjustable keyboard, adjustable keyboard, flat keyboard; (b) flat keyboard, adjustable keyboard, adjustable keyboard; and (c) adjustable keyboard, flat keyboard, adjustable keyboard. Half the participants used a wrist rest and half did not; one-third of them took part in each sequence.

The two within-subjects variables were keyboard day and hour. The three levels of keyboard day were flat keyboard day, SA keyboard Day 1, and SA keyboard Day 2. The four levels of hour were Hours 1 through 4. Each participant was exposed to both keyboards - the SA keyboard for 8 h (4 h on two days) and the flat keyboard for 4 h. The order of keyboard use was balanced so that six participants used the SA keyboard on Days 2 and 3 and the flat keyboard on Day 4, six used the SA keyboard on Days 2 and 4 and the flat keyboard on Day 3, and six used the flat keyboard on Day 2 and the SA keyboard on Days 3 and 4. Participants were randomly assigned to the presentation orders. The dependent variables were software measures of performance (i.e., the number of keystrokes minus the number of backspaces), participant ratings of musculoskeletal discomfort, and experimenter ratings of posture.

Only one fixed configuration of the SA keyboard was used during the three middle days of testing, because we believed that allowing each participant to set the angle would potentially yield too much variability in the angular characteristics for proper comparison. We recognize that fixing the configuration limited a major strength of the SA keyboard- its multiple-axis adjustability - but it still allowed the examination of postural differences that occur when using a split keyboard rather than the flat keyboard.

Equipment and Materials

Two workstations, which had exactly the same components and configuration, were used so that two participants could be tested simultaneously. Each workstation had an IBM PC-AT computer with Microsoft Word word-processing software, which was used for a word-processing text entry task. The PC-AT had either a flat Northgate keyboard or an adjustable Comfort Keyboard, and either a wrist rest or no wrist rest.

The workstations (Wrightline) had a primary working surface that was adjustable in height. Chairs (Steelcase Sensor[TM]) with seat pan height adjustment, an adjustable range of backrest motion, and armrests were used. Each workstation had a document holder on an adjustable arm. One video camera was positioned at each workstation to record the participant's wrist and shoulder postures from the left side.

The flat keyboard had a height of 3 cm at the home row and a 9 [degrees] angle from the horizontal plane. The SA keyboard was put in a fixed configuration that the experimenters believed would reduce hand pronation [ILLUSTRATION FOR FIGURE 1 OMITTED]. The numeric keypad was removed from the SA base. The two halves of the keyboard were then adjusted so that the inside front corners were 9.0 cm apart and the inside back corners were 4.5 cm apart. The inside front corners were then raised 8.0 cm from the base, and the inside back corners were raised 12.5 cm from the base. The outside front corners were raised 1.5 cm from the base. All measurements of height were made from the base to the protruding edge of the keyboard halves. This protruding edge is 0.5 cm from the bottom edge of the keyboard and 1.4 cm from the top edge of the keyboard. This edge was used for measurements because it was easily distinguishable.

Electronic performance monitoring software was used to count keystrokes during typing sessions. The software counted the total number of keystrokes per 1-min period, along with the number of delete and backspace keystrokes. Participants typed a hard-copy document extracted from a novel on technological innovation, which they were unlikely to have previously read. The text was presented to participants as a manuscript typed on 8 1/2 x 11-inch (21.6 x 27.9 cm) sheets of white bond paper.

Participants filled out a questionnaire, developed from the National Institute for Occupational Safety and Health (NIOSH) Health Complaints Questionnaire (Smith et al., 1981), that examined the extent of musculoskeletal pain, psychological distress, and satisfaction with task characteristics. Each item was rated on a four-point scale, with 1 indicating no complaint or pain and 4 indicating having the complaint or pain all of the time. Participants also completed a body pain chart (Kuorinka et al., 1987), which indicated the severity of the pain of any identified body part on a seven-point scale, with 0 indicating no pain and 6 indicating a great deal of pain. At the end of the fifth day of the experiment, each participant was given a semistructured interview to assess reactions to the study and the keyboards. One participant's questionnaire data were lost.


Posture analysis. The postures of the wrists and left shoulder were recorded using a video camera within view of the participant. The postures were reviewed on a videotape recorder that allowed stop-action viewing. The first 40 min of Hour 1, the middle 40 min of Hours 2 and 3 (i.e., the camera was turned on 10 min after the hour began and stopped 10 min before the third session ended), and the last 40 min of Hour 4 were recorded. The postural evaluations rated the extent of wrist pronation, wrist extension/flexion, wrist radial/ulnar deviation, and left shoulder abduction/adduction.

The video camera was positioned to the left of each participant to record shoulder abduction/adduction, wrist flexion/extension, and wrist ulnar/radial deviation. The angle of the camera provided a good characterization of the left hand ulnar/radial deviation and right hand extension/flexion. The examination of posture for left hand extension/flexion and right hand ulnar/radial deviation was not as accurate, given the camera placement and angle, but was adequate for the relative evaluation of the two keyboards and the use of a wrist rest. Pronation of the left hand and wrist was well characterized by this camera position, but examination of right hand pronation was not as accurate as for the left. However, the characterization for the right hand pronation was adequate for relative evaluation.

One 3-min section was evaluated from each of the four recordings of each hour on each day. Thus there were four characterizations of posture of each wrist and left shoulder per participant per day. The 3-min tape section that was reviewed was randomly selected from the recordings for each hour of typing starting with the first hour. For each 3-min section evaluated, the predominant posture for each wrist and each shoulder was scored on a five-point Likert-type scale, with 0 being no deviation, 2 being slight deviation, and 4 being substantial deviation from the neutral posture. This provided a single measure of severity of posture for each wrist and shoulder per 3-min period examined.

The qualitative ordinal scale for the posture ratings provided a means for making relative comparisons. Different random 3-min sections were used to examine the extent of wrist pronation as opposed to those used to examine the other postural factors. By reviewing the videotapes, we determined that once a participant started typing, the general posture of the wrists, shoulders, and back changed very little during the course of a typing session. In addition, the specific postures during an entire 3-min period were consistent. The use of one characterization of posture in a 3-min period appeared to accurately represent the posture of the wrists and shoulders for the entire time.

Two observers conducted the posture-rating task. First, they determined the absolute values for each deviation. With respect to pronation, for example, when the hands were completely face down (fully pronated), a score of 4 (substantial deviation) would be given. If the hands were perpendicular to the horizontal plane, a score of 0 would be given (no deviation). Second, the two observers together rated postures for some time. This procedure was used to standardize the rating system and to obtain agreement as to the rating scale criteria. The observers then rated postures individually, after which they compared their ratings to make sure that they were rating postures consistently.

Day 1. The first day of the experiment consisted of the typing test (performed on the flat keyboard), measurements and adjustments of the workstation, and three 1-h work sessions with a 10-min rest break between sessions. Participants spent 1 hr on the flat keyboard and 2 h on the SA keyboard. They were told to type as quickly but as accurately as possible while ignoring formatting (e.g., boldface or italics). They were also told not to have caffeine or nicotine before or during the work sessions and not to type outside of the experiment during the week of the experiment.

After a participant passed the typing test, the researcher adjusted the workstation to accommodate the participant's physical dimensions (i.e., so that feet were flat on the floor and elbows bent 90 [degrees]). The heights of the chairs and work surface were adjusted and recorded for both the SA keyboard and flat keyboard. (Note: The SA keyboard work surface was lower than that of the flat keyboard because the SA keyboard is higher than a conventional flat keyboard when it is put in an angled configuration.)

Days 2-4. The next three days of the experiment each consisted of four consecutive 1-h work sessions separated by 10-min rest breaks. Participants used the flat keyboard on one of these days and the SA keyboard on the other two days. Participants were randomly assigned to keyboard sequence patterns. They filled out a body pain chart and health questionnaire before Hour 1 (pretest) and after Hour 4 (posttest) each day.

Day 5. The last day of the experiment was used to get feedback from the participants on different configurations of the SA keyboard. The results of this day are not reported here.



Performance data were collected continuously during the four 1-h work sessions by monitoring software for Days 2 through 4. The data were grouped into 1-min records for each hour. The 1-min records were aggregated into 10-min and 60-min intervals for analysis. Multiple analyses of variance (MANOVAs) were performed for the 60-min interval summaries, with the time intervals acting as the dependent variables. For all reported performance means, the measure used is adjusted keystrokes. The use of MANOVA for repeated-measures analyses is preferred to univariate repeated measures because the former holds no assumptions regarding sphericity (Davidson, 1972; McCall & Appelbaum, 1973). Mauchly's test for sphericity did reveal violations of the assumption.

The results of the Wrist Rest x Sequence x Keyboard Day x Hour MANOVA showed that when collapsing over all three keyboarding days, there was a significant main effect for hour, F(3, 10) = 15.820, p [less than] .05, and there was also a significant linear trend, F(1, 12) = 49.987, p [less than] .05, as shown in Figure 2. There were no wrist rest or sequence effects, nor was there any performance difference among the three days.

When the MANOVA was used to compare the first and second days of using the SA keyboard, a significant Wrist Rest x Day of Use interaction was found, F(1, 12) = 5.429, p [less than] .05 [ILLUSTRATION FOR FIGURE 3 OMITTED]. When a wrist rest was not used, performance on the first and second clays of using the SA keyboard was very similar (SA Day 1 M = 16834, SD = 2385, and SA Day 2 M = 16858, SD = 2118). However, when a wrist rest was used, performance on the second day of using the SA keyboard was considerably higher (SA Day 1 M = 17272, SD = 1402, and SA Day 2 M = 18006, SD = 1475). This interaction explains the main effect for day of use, F(1, 12) = 6.207, p [less than] .05, when the mean for the first day of using the SA was 17053 (SD = 1911) adjusted keystrokes and the mean for the second day was 17432 (SD = 1866) adjusted keystrokes. Also, as before, there was a significant main effect for hour, F(3, 10) = 11.479, p [less than] .05, with a corresponding significant linear trend F(1, 12) = 31.639, p [less than] .05. Again, the means show a decline over the day (see Table 1).

Flat keyboard use was also compared with the average of both days of using the split, adjustable keyboard. The only significant effect was for hour, F(3, 10) = 14.604, p [less than] .05, and its corresponding significant linear trend, F(1, 12) = 49.041, p [less than] .05. The means show a decline in performance over time (see Table 1).


The Wilcoxon signed-ranks test was used to analyze differences between keyboards, and the Mann-Whitney test was used to examine differences between wrist rest use and nonuse. Nonparametric tests were used instead of the more common parametric tests because the dependent variables were not normally distributed. For analyses between the keyboards, the ratings for the two days using the SA keyboard were averaged. In all of the analyses, the ratings were averaged over the 4 h. This yielded a single rating for each posture with each type of keyboard. The ratings ranged from 0 to 4, which corresponded to no deviation and substantial deviation, respectively. The ratings provided a means for relative comparison.

There were higher ratings (p [less than] .05) for left hand extension, right hand extension, and left hand radial deviation for participants using the SA keyboard than for those using the flat keyboard. There were higher ratings (p [less than] .05) for left hand ulnar deviation, right hand ulnar deviation, and pronation for participants using the flat keyboard than for those using the SA keyboard. More left hand extension was found for participants who did not use a wrist rest (p [less than] .05). All mean ratings, except for pronation, ranged from 0.0 to 1.2, indicating that the extent of postural deviations was, at most, slight. The mean rating for pronation was 3.94 (substantial) while using the flat keyboard and 2.33 (moderate) while using the SA keyboard.

Musculoskeletal Pain

The 17 participants' responses to the ratings on the NIOSH health checklist questionnaires, which had a scale of 1 (not at all) to 4 (very much pain), and the body charts, which used a scale of 0 (no pain) to 6 (severe pain), were evaluated using Mann-Whitney and Wilcoxon signed-rank tests. The former was used to evaluate differences between independent observations (specifically, those participants using wrist rests versus those not using wrist rests), whereas the latter was used to evaluate all repeated-measures tests (e.g., pretest vs. posttest). Nonparametric tests were used instead of the more common parametric tests because the dependent variables were [TABULAR DATA FOR TABLE 1 OMITTED] not normally distributed. Tables 2 and 5 show only the eight questions from the NIOSH health checklist that relate to musculoskeletal discomfort or tiredness.

Keyboard. Comparisons between the single day of flat keyboard typing and the average of the two days of typing on the SA keyboard showed no differences between the pretests and the posttests for musculoskeletal pain. Similarly, the comparisons between the single day of flat keyboard typing and each of the two days of typing on the SA keyboard separately showed no differences in the reporting of pain for either the NIOSH health checklist or the body pain chart.

When participants used the flat keyboard (see Table 2), there were significant differences in responses to the NIOSH checklist between the pretests and posttests for hand and finger pain, back pain, wrist and hand pain, arm pain or numbness, pain in the shoulder or neck, and low back pain. When they used the flat keyboard there were also significant differences in responses to the body pain chart between the pretests and posttests for neck, back, inside elbow and forearm, outside elbow and forearm, palmar left hand, palmar right hand, dorsal left hand, and dorsal right hand.

Significant differences were also reported between pretests and posttests for the NIOSH checklist when participants used the SA keyboard the first day. These differences were for hand and finger pain, back pain, wrist and hand pain, arm pain or numbness, tiredness, pain in the shoulder or neck, and low back pain. The pretest/posttest differences reported on the first day using the SA keyboard for the body pain chart were for the neck, back, back of shoulder, inside elbow and forearm, dorsal side of wrist, palmar side of wrist, and dorsal side of the left hand (see Table 2). Table 2 also shows the NIOSH checklist and body pain chart significant differences between the pretests and posttests for the second day using the SA keyboard.

Wrist rest. Three significant differences (p [less than] .05) in musculoskeletal pain were found between the posttests of those who used wrist rests and those who did not. Participants who used wrists rests (M = 3.06, SD = 0.62) reported feeling more in control of their typing than did those who did not use wrist rests (M = 2.25, SD = 0.58). Also, participants who did not use a wrist rest reported more pain in the front of their shoulders (M = 1.17, SD = 1.57 vs. M = 0.19, SD = 0.38) and more pain at the outside of their elbow/forearm (M = 1.35, SD = 1.36 vs. M = 0.19, SD = 0.56) than did those who did use a wrist rest.

Significant differences were also found between the pretests and posttests for both wrist rest users and nonusers. For the former group (see Table 3), the differences on the NIOSH checklist were for hand and finger pain, back pain, wrist and hand pain, pain in the shoulder or neck, and low back pain. The differences for the body pain chart were for the neck, back, inside elbow and forearm, palmar side of wrists, palmar side of left hand, and dorsal side of left hand. For the wrist rest nonusers (see Table 3), the responses differed on the NIOSH checklist for hand and finger pain, back pain, wrist and hand pain, arm pain or numbness, tiredness, pain in the shoulder or neck, liking the job, and low back pain. The significant body pain chart differences were at the neck, back, front shoulder, inside elbow and forearm, outside elbow and forearm, dorsal side of wrist, palmar side of wrist, dorsal side of left hand, and dorsal side of right hand.


The purpose of this study was to examine the capability of experienced typists to perform [TABULAR DATA FOR TABLE 2 OMITTED] text typing on the SA keyboard as compared with a typical standard flat keyboard with a similar key layout and key action characteristics. The key objective was to determine whether the SA keyboard could promote better typing posture for the shoulders and wrists and reduce musculoskeletal complaints without hindering performance.

The SA keyboard is designed to allow for numerous configurations for each hand. For the purposes of experimentation, however, we believed it wood be more manageable, in terms of the treatment exposure and data analysis, to limit the configuration to a single setup that was considered to be in conformance with accepted ergonomics principles for wrist and shoulder postures (Grandjean, 1988). Previous research on experimental keyboards with characteristics substantially different from those of a traditional flat keyboard have shown some positive benefits for upper limb, wrist posture, and musculoskeletal risk of injury (Nakaseko et al., 1985; Rempel, Gerson, Armstrong, Foulke, & Martin, 1991; Thompson et al., 1990). However, [TABULAR DATA FOR TABLE 3 OMITTED] at least one study's results have shown that experienced typists were not able to adapt to the new motor patterns required by the use of an experimental keyboard and, consequently, were not able to perform as well as with a traditional keyboard (Hertting-Thomasius, Steidel, Prokop, & Lettow, 1992).

In contrast, our study indicates that experienced typists were able to perform as well, in terms of quantity and quality of typing text materials, with the SA keyboard as with the traditional flat keyboard. These results were obtained with very little practice (about 2 h) and no familiarity with the SA keyboard before the experiment began. In fact, participants' ability to use the SA keyboard effectively was apparent from the start of the experiment. This may have been because the SA keyboard has keying characteristics, such as keying force and keyboard layout (QWERTY), that are similar to those of many standard keyboards, such as the flat keyboard used in this study. Swanson, Galinsky, Steward, and Pan (1994) also discovered that performance was the same whether using a split keyboard design or a flat keyboard. In their study there were no differences between keyboards in errors per hour, and the difference in typing speed had vanished by the third day of the study. Cakir (1995) and Swanson et al. (1997) found similar performance results.

The ability of experiment participants to rapidly adapt to the SA keyboard is good news for those who believe that alternative keyboards hold promise for reducing the risk factors for upper extremity trauma caused by typing tasks. This short adaptation period with regard to performance on the SA keyboard suggests that experienced typists will have little difficulty in transitioning to such experimental keyboards that retain the traditional keyboard "feel" and layout. Feel was defined by Chen et al. (1994) as subjective evaluations of keyboard touch, keying rhythm, effort, and awkwardness. Their results showed that people prefer the feel of split keyboard designs over that of flat keyboards. Likewise, Cakir (1995) found that when comparing a traditional keyboard with an adjustable keyboard, participants found no differences in four of six "familiarity" ratings.

The characteristics of the task requirements in our study had a strong influence on participants' performance. They typed text continuously for 60 min without an opportunity to take a break or to change to other tasks that would allow for different postures and motion patterns. These conditions were similar to those in the Swanson et al. (1997) study, in which participants typed continuously for 75-min periods. In their experiment there were significant work period effects for discomfort and fatigue, showing that the two measures increased throughout the workday. The task used in the study presented in this paper was intensive in terms of concentration, typing demands, and fixed body postures. We believe that this demanding and stressful situation was the cause of the decline in typing performance that occurred between the first and fourth hours each day, regardless of which keyboard was being used. This belief is based on the claim that stress may have adverse affects on performance (Hancock, Chignell, & Vercruyssen, 1990; Smith, 1987).

The fact that the participants were in workstations that had been adjusted to their specific anthropometric characteristics is particularly interesting. This is important because it may illustrate that the workload and content aspects of a typing task can have at least as great a role in influencing typists' performance as can the physical workstation and interface design characteristics. This may be an interesting direction for future studies. In this study high levels of performance could not be sustained even with the positive physical qualities of the workstation and the adjustable keyboard. The heavy demands of typing text continuously for 1 h, without a break, produced performance deficits within each work session and over work sessions on the same day, even with "good" ergonomic furniture and proper adjustments.

In addition to the performance findings comparing the SA keyboard with the standard flat keyboard, there was also an interesting finding with regard to use of a wrist rest. When we compared the first and second days of SA keyboard use, we discovered that the participants who did not use a wrist rest had no change in performance but that when a wrist rest was used, performance increased from the first to the second day. This may have resulted from the increased comfort that wrist rest users reported in their shoulders and forearms. Other explanations are also possible; Miller, Straker, and Pollock (1994) found no relationship between discomfort and typing speed.

Given the task's influence on performance over the course of the day, as found in the study presented in this paper, it is not surprising that there was also an increase in reported musculoskeletal pain over the course of the day. Participants reported more neck, shoulder, back, arm, elbow, wrist, and hand pain over the course of the day of typing when using either the SA keyboard or the flat keyboard. There were no differences between keyboards, though, in the extent of pain reported by participants. Two recent studies also found no differences in musculoskeletal pain when comparing a traditional keyboard with an adjustable keyboard. However, both found differences from the beginning to the end of the sessions (Cakir, 1995; Swanson et al., 1997). Again, this illustrates that even when workstation and keyboard characteristics have "good" ergonomic features, the task characteristics can overpower their benefits and cause typists to experience musculoskeletal pain and reduced performance.

There were two interesting effects of the wrist rest with regard to pain complaints. Participants who did not use a wrist rest indicated on the body pain chart that they experienced more pain in their elbow or forearm and in the front of the shoulder than did wrist rest users. The increased shoulder pain reported in the absences of a wrist rest may be attributable to the fact that the shoulders support the weight of the arms in the absence of a support such as a wrist rest.

The SA keyboard resulted in both postural benefits and postural disappointments. All of the postural differences that existed between keyboards had means between 0.0 and 1.2 on a scale anchored at 4. This indicates that despite the statistical differences, there may be no practical value to the postural differences. This may have occurred because the participants were not allowed to configure the SA keyboard to their own preferences. The difference between keyboards for pronation was more substantial, however. The fact that the SA keyboard configuration was constrained to one fixed position may have limited the extent of pronation reduction achievable for individual participants. The only postural difference between wrist rest users and nonusers was that the former had decreased left hand extension.

We believe that letting participants establish their own configuration may eliminate the postural situations in which the SA keyboard produced slightly greater wrist deviation than the flat keyboard. Burastero et al. (1994) in fact found that by allowing participants to adjust split keyboards to a configuration they found most comfortable, wrist ulnar deviation and extension were less than that obtained with use of a standard fiat keyboard.

This study supports results of prior research showing that there are no performance differences between traditional and split, adjustable keyboards. Furthermore, it seems that work duration is more important than keyboard design with regard to musculoskeletal pain. Posture was affected by keyboard design, but the only practical difference may be that split, adjustable keyboards offer the potential for reduced pronation.


The authors express their appreciation to the Health Care Keyboard Company, Inc., Menomonee Falls, Wisconsin, for providing the Comfort[TM] Keyboard prototype that was used in this research.


Bammer, G., & Martin, B. (1988). The arguments about RSI: An examination. Community Health Studies, 12, 348-358.

Bammer, G., & Martin, B. (1992). Repetitive strain injury in Australia: Medical knowledge, social movement, and de facto partisanship. Social Problems, 39, 219-237.

Bergqvist, U., Wolgast, E., Nilsson, B., & Voss, M. (1995). Musculoskeletal disorders among visual display terminal workers: Individual, ergonomic, and work organizational factors. Ergonomics, 38, 763-776.

Bernard, B., Sauter, S., Peterson, M., & Hales, T (1992). Upper extremity evaluation of musculoskeletal outcomes among newspaper employees. In H. Luczak, A. Cakir, & G. Cakir (Eds.), Abstracts book: Work with display units WWDU '92 (p. D-15). Berlin: Technische Universitat Berlin.

Bongers, P. M., de Winter, C. R., Kompier, M. A. J., & Hildebrant, V. H. (1993). Psychosocial factors at work and musculoskeletal disease. Scandinavian Journal of Work Environment and Health, 19, 297-312.

Burastero, S., Tittiranonda, R, Chen, C., Shih, M., & Rempel, D. (1994). Ergonomic evaluation of alternative computer keyboards. In A. Grieco, G. Molteni, E. Occhipinti, & B. Piccoli (Eds.), Work with display units WWDU '94 (Vol. 3, pp. C15-16). Milan: University of Milan.

Cakir, A. (1994). Do split keyboards help to reduce strain? In A. Grieco, G. Molteni, E. Occhipinti, & B. Piccoli (Eds.), Work with display units WWDU '94, (Vol. 3, p. C-24). Milan: University of Milan.

Cakir, A. (1995). Acceptance of the adjustable keyboard. Ergonomics, 38, 1728-1744.

Cakir, A., Hart, D. J., & Stewart, T. F. M. (1979). The VDT manual. New York: Wiley.

Chen, C., Burastero, S., Tittiranonda, P., Hollerbach, K., Shih, M., & Denhoy, R. (1994). Quantitative evaluation of four computer keyboards: Wrist posture and typing performance. In Proceedings of the Human Factors and Ergonomics Society 38th Annual Meeting (pp. 1094-1098). Santa Monica, CA: Human Factors and Ergonomics Society.

Davidson, M. L. (1972). Univariate versus multivariate tests in repeated measures experiments. Psychological Bulletin, 77, 446-452.

Grandjean, E. (1979). Ergonomical and medical aspects of cathode ray tube displays. Zurich: Federal Institute of Technology.

Grandjean, E. (1988). Fitting the task to the man: A textbook of occupational ergonomics (4th ed.). London: Taylor & Francis.

Gunnarsson, E., & Ostberg, O. (1977). The physical and psychological working environment in a terminal-based computer storage and retrieval system (NBOSH Report 35). Stockholm: National Board of Occupational Safety and Health.

Hagberg, M. (1981). Workload and fatigue in repetitive arm elevations. Ergonomics, 24, 543-555.

Hales, T., Sauter, S, Peterson, M., & Bernard, B. (1992). Upper extremity musculoskeletal conditions among VDT users in a telecommunications company. In H. Luczak, A. Cakir, & G. Cakir (Eds.), Abstracts book: Work with display units WWDU '92 (p. D-14). Berlin: Technische Universitat Berlin.

Hancock, P. A., Chignell, M. H., & Vercruyssen, M. (1990). Predicting the effects of stress on performance. In Proceedings of the Human Factors Society 34th Annual Meeting (pp. 1081-1085). Santa Monica, CA: Human Factors and Ergonomics Society.

Hertting-Thomasius, R., Steidel, F., Prokop, M., & Lettow, H. (1992). On the introduction of ergonomically designed keyboards. In H. Luczak, A. Cakir, & G. Cakir (Eds.), Abstracts book: Work with display units WWDU '92 (p. P-2). Berlin: Technische Universitat Berlin.

Hunting, W., Laubli, T., & Grandjean, E. (1981). Postural and visual loads at VDT workplaces. Ergonomics, 24, 917-931.

Kuorinka, L, Jonsson, B., Kilbom, A., Vinterberg, H., Biering-Sorensen, F., Andersson, G., & Jorgensen, K. (1987). Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Applied Ergonomics, 18, 233-237.

Kroemer, K. H. E. (1972). Human engineering the keyboard. Human Factors, 14, 51-63.

LeGrande, D. (1992). VDT repetitive motion health concerns in the U.S. telecommunications industry. In H. Luczak, A. Cakir, & G. Cakir (Eds.), Abstracts book: Work with display units WWDU '92 (p. D-43). Berlin: Technische Universitat Berlin.

LeGrande, D., & Eisen, D. (1988, August). Upper extremity cumulative trauma disorders in VDT users. Paper presented at the Promoting Health & Productivity in the Computerized Office: Models of Successful Ergonomic Interventions Conference, Oxford, Ohio.

Mandal A. C. (1982). The correct height of school furniture. Human Factors, 24, 257-269.

Marek, T., Noworol, C., Wos, H., Karwowski, W., & Hamiga, K. (1992). Muscular loading and subjective ratings of muscular tension by novices when typing with standard and split-design computer keyboards. International Journal of Human-Computer Interaction, 4, 387-394.

McCall, R., & Appelbaum, M. (1973). Bias in the analysis of repeated-measures designs: Some alternative approaches. Child Development, 44, 401-415.

Miller, J., Straker, L., & Pollock, C. (1994). The relationship between discomfort and typing performance of a VDU task in different postures. In A. Grieco, G. Molteni, E. Occhipinti, & B. Piccoli (Eds.), Work with display units WWDU '94 (Vol. 3, pp. C17-19). Milan: University of Milan.

Nakaseko, M., Grandjean, E., Hunting, W., & Gierer, R. (1985). Studies of ergonomically designed alphanumeric keyboards. Human Factors, 27, 175-187.

Ohlsson, K., Attewell, R. G., Johnsson, B., Ahlm, A., & Skerfving, S. (1994). An assessment of neck and upper extremity disorders by questionnaire and clinical examination. Ergonomics, 37, 891-897.

Powers, J. R., Hedge, A., & Martin, M. G. (1992). Effects of full motion forearm supports and a negative slope keyboard support system on hand-wrist posture while keyboarding. In Proceedings of the Human Factors Society 36th Annual Meeting (pp. 796-800). Santa Monica, CA: Human Factors and Ergonomics Society.

Rempel, D., Gerson, l., Armstrong, T., Foulke, J., & Martin, B. (1991). Fingertip forces while using three different keyboards. In Proceedings of the Human Factors Society 35th Annual Meeting (pp. 253-255). Santa Monica, CA: Human Factors and Ergonomics Society.

Rempel, D., & Horie, S. (1994). Effect of wrist posture during typing on carpal tunnel pressure. tn A. Grieco, G. Molteni, E. Occhipinti, & B. Piccoli (Eds.), Work with display units WWDU '94 (Vol. 3, pp. C27-28). Milan: University of Milan.

Rose, M. J. (1991). Keyboard operating posture and actuation force: Implications for muscle over-use. Applied Ergonomics, 22, 198-203.

Smith, M. J. (1984). Health issues in VDT work. In J. Bennett, D. Case, J. Sandelin, & M. J. Smith (Eds.), Video display terminals: Usability issues and health concerns (pp. 195-228). Englewood Cliffs, NJ: Prentice Hall.

Smith, M. J. (1987). Occupational stress. In G. Salvendy (Ed.), Handbook of human factors engineering (pp. 844-860). New York: Wiley.

Smith, M. J., Stammerjohn, L, Cohen, B. G. F., & Happ, A. (1981). An investigation of health complaints and job stress in video display operations. Human Factors, 23, 387-400.

Sommerich, C. M. (1994). Carpal tunnel pressure during typing: Effects of wrist posture and typing speed. In Proceedings of the Human Factors and Ergonomics Society 38th Annual Meeting (pp. 611-615). Santa Monica, CA: Human Factors and Ergonomics Society.

Swanson, N. G., Galinsky, T. L., Cole, L. L., Pan, C. S., & Sauter, S. L. (1997). The impact of keyboard design on comfort and productivity in a text-entry task. Applied Ergonomics, 28, 9-16.

Swanson, N. G., Galinsky, T. L., Steward, L. L., & Pan, C. S. (1994). The effects of keyboard design features on comfort and productivity. In A. Grieco, G. Molteni, E. Occhipinti, & B. Piccoli (Eds.), Work with display units WWDU '94 (Vol. 3, pp. C13-14). Milan: University of Milan.

Szabo, R. M., & Chidgey, L. K. (1989). Stress carpal tunnel pressures in patients with carpal tunnel syndrome and normal patients. Journal of Hand Surgery, 14, 624-627.

Thompson, D. A., Thomas, J., Cone, J., Daponte, A., & Markison, R. (1990). Analysis of the TONY! variable geometry keyboard. [n Proceedings of the Human Factors Society 34th Annual Meeting (pp. 365-369). Santa Monica, CA: Human Factors and Ergonomics Society.

Michael J. Smith is chair of the Department of Industrial Engineering at the University of Wisconsin-Madison. He received his Ph.D. in industrial psychology in 1973 from the University of Wisconsin-Madison.

Ben-Tzion Karsh is a Ph.D. candidate in industrial engineering at the University of Wisconsin-Madison, where he received his M.S. in industrial engineering in 1996.

Frank T. Conway is a National Research Council postdoctoral associate at the National Institute for Occupational Safety and Health Robert A. Taft Laboratories. He received his Ph.D. in industrial engineering in 1996 from the University of Wisconsin-Madison.

William J. Cohen is a professor at the Business School of Massey University in Palmerston North, New Zealand. He received his Ph.D. in industrial engineering in 1997 from the University of Wisconsin-Madison.

Craig A. James is a staff research analyst at Allstate Insurance. He received his Ph.D. in industrial engineering and psychology in 1997 from the University of Wisconsin-Madison.

Jay J. Morgan is a process engineer at CUNA Mutual Insurance Company. He received his M.S. in industrial engineering in 1993 from the University of Wisconsin-Madison.

Katherine Sanders is an associate scientist for the Wisconsin Center for Education Research at the University of Wisconsin-Madison, where she received her Ph.D. in industrial engineering in 1993.

David J. Zehel is a Ph.D. candidate in industrial engineering at the University of Wisconsin-Madison, where he received his M.S. in industrial engineering in 1993.
COPYRIGHT 1998 Human Factors and Ergonomics Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Smith, Michael J.; Karsh, Ben-Tzion; Conway, Frank T.; Cohen, William J.; James, Craig A.; Morgan, J
Publication:Human Factors
Date:Jun 1, 1998
Previous Article:Principles of educational multimedia user interface design.
Next Article:Workplace changes associated with a reduction in musculoskeletal symptoms in office workers.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters