Assistive Technology Research Institute
College Misericordia - Dallas, PA 18612
 
Founded and Sponsored by the Sisters of Mercy of Dallas

 

Morse Vrs Headpointer

 



Long-term Speed and Accuracy of Morse code vs. Head-pointer Interface for Text Generation

Denis K. Anson, MS, OTR/L, Melinda Glodek, OTS, Richard M. Peiffer, OTS, Cory G. Rubino, OTS, Patrick T. Schwartz, OTS

Please address all correspondence to the first author at danson@misericordia.edu .

Abstract

Objective. While there is some anecdotal evidence regarding the long-term usefulness of Morse code vs. head- pointer interfaces, it is seldom compelling and often conflicting. The purpose of this study is to determine which of two alternative computer input systems produces a greater rate of output once proficiency is reached.

Method. Participants were 8 able-bodied people, ranging in age from 19 years to 59 years old. Four participants had prior experience with Morse code, while none had experience with head-pointers and on-screen keyboards. A single-subject, successive interventions design was employed to measure individual performance for both input methods.

Results. All users with no prior Morse code experience were able to type faster with the combination of head-pointing and on-screen keyboard based on short term trials. All 4 participants with Morse code experience reached a greater input speed with Morse code. There was no systematic difference in accuracy observed between the two input methods.

Conclusion. Over a short-term assessment, the combination of on-screen keyboard and head-pointer provides input rates that are faster than those achieved by Morse code. However, individuals who have more experience with Morse code were able to type faster with Morse code than with the on-screen keyboard.

Introduction

Disabilities, such as spinal cord injury, MS and Cerebral Palsy can limit an individual's ability to control his/her hands, but spare the ability to move his/her head and face (K. L. Reed, 2001) . For the person with intact movement above the shoulders, but limited movement distally, there are few available methods of functional computer control. Among these is the relatively infrequently used Morse code input and the more commonly used combination of head pointing devices and on-screen keyboards. Each of these input methods has inherent advantages and disadvantages. Morse code input, according to long-term users, eventually becomes automatic, but requires significant training to achieve proficiency. The level of control offered by head-pointers is closely comparable to that of a standard mouse; however, limited endurance and restricted range of motion of the head and neck may limit their use (Anson, 1997) .

Since different access methods make different demands on the person with a disability, the therapist must carefully match the functional ability of the person to the enabling device. According to the Human Interface Assessment Model (HIA), the role of assistive technology is to bridge the gap between the abilities of the individual and the demands of the task (Anson, 2001) . One essential goal of an occupational therapist is to provide a consumer with the means to achieve the highest degree of function possible. For even a person with a severe disability (one which renders the individual dependent in one or more major life tasks) this may include joining or returning to the workforce.

The United States currently has more disabled individuals in the workforce than ever (Mergenhagen, 1997) . This increased employment of individuals with disabilities may derive from a combination of factors. Improving medical technologies are saving the lives of individuals with severe disabilities who, in earlier times, would not have survived their injuries. Improved assistive technologies are giving these people the ability to control their environments and to produce meaningful work. Finally, the Americans with Disabilities Act of 1990 ("Americans with Disabilities Act of 1990," 1990) has given people with severe disabilities legal protections in their efforts to become active members of society.

In order to support the efforts of the client with a disability to return to work, a therapist must stay abreast of the assistive technology devices that would provide the highest levels of efficiency and productivity. While the individual with a disability may always have some performance lag as compared to their able-bodied co-workers, it is incumbent upon the clinician to minimize any performance limitations by providing the best possible access method.

Unfortunately, most clinical decisions between different access technologies must be based on short-term clinical trials. While there is some anecdotal evidence regarding the long-term utility of alternative devices, it is seldom compelling and often conflicting. Such cases tend to focus on extreme performance of "star" users, and do not reflect performance of the typical individual with a disability. Comparative information on the long-term utility of alternative access technologies would be especially valuable to the clinician making equipment recommendations, and to the person with a severe disability who wants to rejoin the workforce. The purpose of this study is to determine which of two alternative computer input systems, Morse code or head-pointing with an on-screen keyboard, produces a greater rate of output once proficiency is reached.

Background

Morse code

When employed with alternative access computer systems, the Morse characters are transmitted through micro-switches to a hardware or software keyboard emulator, which translates them into orthographic symbols that produce text or speech output (Anson, 1997) . One substantial advantage of Morse code reported by users is that it can eventually become as automatic to the user as touch-typing. According to King (King, 2000) , ".motor, auditory, and visual components of code entry all become automatic; that is, they become sub-cognitive, done without thought" (p.113). Experienced Morse code users commonly report that they do not know the codes for each of the letters, just as touch-typists are not consciously aware of the position of individual letters on the keyboard (Anson, 1997) . The physical demands of Morse code can be as small as a barely palpable muscle twitch, which makes it useful for individuals with profound limitations in endurance. Further, Jarus observed that Morse code ".does not require constant visual contact with the control display and, therefore, can be suitable for a vast number of people" (Jarus, 1994) .

One early article concerning Morse code as a means of alternative communication for persons with disabilities appeared in the September- October 1961 issue of Cerebral Palsy Review (Clement, 1961) . Clement, a speech therapist, described how she had taught three children with cerebral palsy to use Morse code as a method of communication. Over a three-month period, she taught each child to perform an array of facial and extremity movements to convey "dits" and "dahs" while their communication partner would interpret their movements with the help of a Morse code chart. As a result of this training, each of her three clients was able to actively communicate thoughts and ideas with family and friends. A disadvantage noted for this technique was that the code had to be interpreted by another person (McDonald, Schwejda, Marriner, Wilson, & Ross, 1982) . Thus, an individual conversing with one of the children would need to take a moment to interpret the code and then respond, which would impede the fluency of the conversation. However, with the availability of Morse code devices that interface with computers, this disadvantage has been overcome. The code is electronically translated into alphabetical characters that appear on the monitor screen.

A study conducted by McDonald et al. (McDonald et al., 1982) , compared the rate of text output between Morse and scanning for the purpose of developing an Alternative Communication System (ACS) and computer input. The participants were five school-aged children with similar physical disabilities that limited their abilities to speak and access computers. In this study, Morse code was found to have advantages over scanning. First, it was found that four out of five participants produced output faster with Morse code than with scanning. Second, Morse code produced auditory feedback and required no visual interpretation. Thus, the user did not have to look back and forth from a source document to a video screen, which was required when using scanning input. Finally, Morse code can be considered a direct selection method. Rather than wait for the desired group and letter to be highlighted, the user is able to perform actions that will immediately generate the desired character or word. While on-screen keyboards do allow direct selection, they also require visual attention. Hence, the "perceptual advantage" for Morse code found in the McDonald et al study (1982) suggests a similar advantage for Morse code as compared with on-screen keyboards.

Morse code appears to offer one of the fastest available means of text generation for clients with disabilities. In a case study reported by Beukelman, Yorkston, and Dowden (Beukelman, Yourkston, & Dowden, 1985) after six weeks of training for approximately an hour per day, an individual with a C-4 spinal cord injury was able to recreate text with 90% accuracy at the rate of 18 words per minute. While one must be careful not to report extreme cases as if they were typical, many anecdotal reports on Morse code show typing speeds in this range.

Head-pointer Systems

The idea of using head movements to compensate for lost hand function as a means of typing and interfacing with computers has been around for quite some time. Early head-pointers consisted of a rod fastened to the user's forehead with a headband. Another low-tech device, the mouth stick, consists of a rod attached to a mouthpiece. The mouth stick offers the potential to use oral movements in conjunction with head and neck movements to type, but also can cause problems ranging from temporal-mandibular joint (TMJ) dysfunction to dental migrations These low-tech mechanical head-pointers are still used, often when manipulation of physical objects is needed in addition to text generation (Anson, 2001) . Mechanical head-pointers are somewhat limited in that users achieve relatively low degrees of accuracy; lose visual fixation on source documents while typing, and achieve relatively slow typing speeds (DeVries, Deitz, & Anson, 1998) . High-technology head-pointers sacrifice the ability to manipulate physical objects in order to achieve higher degrees of accuracy and efficiency when typing on a computer.

One of the first high-technology head pointing systems, the "View Control System" was designed for able-bodied persons as a means for eliminating the need to manually move a mouse. Touch typists were the initial target population because it was believed that keeping both hands free to type at all times would increase productivity and would therefore be attractive to these individuals (Frame, 1994) . Because the demand for head-pointers among touch typists was fairly low, sales of the systems fell far below expectations. The technology was then remarketed to the disabled population and sales increased somewhat. This original system is now marketed, with some improvements in the computer interface, as the HeadMaster Pro and is still sold today.

Mouse emulators that are based on head pointing enable persons with impaired or lost hand function to move the on-screen mouse pointer. Combined with an on-screen keyboard, the system allows a person with head-control and little else to generate text. While the speed of text generation of head-pointers varies in earlier studies (Angelo, Deterding, & Weisman, 1991; DeVries et al., 1998) clinical experience suggests typing speeds on the order of 15 to 20 words per minute are commonly attainable.

Although the anecdotal reports of typing speeds are faster for Morse code users than for head-pointer users, reports from the field show that, while head-pointing and on-screen keyboards are commonly accepted, Morse code is rarely used by people with disabilities. Because clinicians should recommend the system that will provide the best chance for functional performance, objective information on the relative utility of these two input methods is important. This is especially true since many people can use either of these input methods, but have very few alternatives available. For individuals who use these access systems for vocational purposes, a difference in input rates of as small as 2% could result in the equivalent of an extra week of work over the course of a year.

The results of this study may assist the clinician in selecting input technologies that will effectively compensate for decreased physical function and allow people with severe disabilities achieve the highest attainable productivity.

Our research hypothesis is that, on initial training, the headpointing system will allow faster typing, but over time, Morse code typing will be faster. Our second hypothesis is that there will not be a significant difference in accuracy between the input devices.

Methodology

Design

A single-subject, successive intervention design was used to compare the typing speed and accuracy of a head pointing system combined with an on-screen keyboard and Morse code text input.

Assumptions

In order to provide a consistent task, all typing in this study consisted of copying from provided text. Although much real-world typing is confounded with composition, the authors assume that the efficacy of typing from copy will reflect typing during composition. This assumption ignores the effect that different levels of cognitive overhead of the two input methods might have on the process of composition. The term "cognitive overhead" describes the amount of effort and concentration required to control the process, reducing the amount of concentration available to produce the product. If one input method is significantly more demanding, it might increase, or even reverse the differences found in this study. Since we do not feel that these two input methods are markedly different in cognitive demands, we feel that the relative merits of the two devices would not change during composition tasks.

Task (Operational Definitions)

Typing speed was indicated by the number of words typed during the twenty-minute trials as counted by the "Word Count" feature of Microsoft Word 2000 . An error was considered to be any discrepancy between the participant's work and the original source document as identified by Microsoft Word 2000's, "Compare Documents" feature. Types of errors included letter or word reversals, dropped letters, missing word, or missing sentences. Any of these or any combination of these in a single block of text was considered to be a single error. While this may result in undercounting of differences, it does assure the possibility of agreement in error counting.

Percent accuracy calculated using the formula: Percentage accuracy = (1-(total errors for passage/total words typed))*100

This results in an accuracy value that is independent of typing speed.

Performance plateau was achieved when the participant produced three trials with typing speeds within 7 percent of each other (Krishef, 1991) . Prior studies conducted by the first author have shown the arbitrarily-selected 7 percent standard to be within normal variation for a stable participant, but not to be achieved while the participants are making consistent gains (Kanny & Anson, 1991) .

Participants

The participants for this study were eight able-bodied individuals from 19 to 59 years old with a mean age of 34.6 years. There were five male and three female participants. All participants were able to read 12-point type (with or without visual correction) that was presented to them in the source document. In addition, all participants were capable of hearing the tones produced by the Morse code device. Half of the participants selected for this study were long-term users of Morse code recruited from a local ham-radio club although none were familiar with sip and puff input. Additionally, none of participants had prior experience with head pointing. This mix was used to evaluate both the short and long term performance for Morse input.

None of the research participants were physically disabled. While this might seem to limit the generalizability of the study, we feel that it does not. The individuals for whom Morse code and headpointing are both viable input methods have normal or nearly normal head movement and oral-motor control. The able-bodied experimental participants are, therefore, reasonable analogues to the disabled individuals they are representing, in terms of head and intra-oral control.

Apparatus

Two personal computers equipped; with Pentium II processors operating at 450 MHz, 128 MB of memory, and 17-inch curved monitors were used in the study. Text was typed into Microsoft Word 2000, which was also used to evaluate the speed and accuracy of typing. Morse code input was through the Darci Card Morse code interface device connected to a sip and puff switch . This switch was delivered with an in-line filter to prevent saliva contamination. We found that the in-line filter slowed the response of the switch to a level that made rapid text generation impossible. Because of this, the in-line filter was removed, and interchangeable suction tips were used for hygiene.

Headpointing input was performed using the HeadMaster Plus in conjunction with ScreenDoors 2000 on-screen keyboard program. The on-screen keyboard was displayed using the "frequency based" Chubon layout, because earlier studies have shown that this keyboard pattern allowed participants to produce faster text entry than QWERTY (Anson, George, Galup, Shea, & Vetter, 2001; Chubon & Hester, 1988) . The word prediction feature of ScreenDoors was disabled during the typing sessions.

The source document used in the study was Joseph Conrad's novel The Secret Agent . The novel was obtained in electronic format through Project Gutenberg . The text document was divided into segments of approximately 500 words (each segment ended at the end of a paragraph which contained the 500 th word.) and was formatted in Microsoft Word 2000 to remove extraneous symbols and line breaks, used 12-point Times Roman font and be double spaced to improve readability.

Procedure

A balanced-order assignment was used to determine the initial input device for each of the participants in order to control for possible order effects. Before beginning the initial typing session with each device, the testers gave a brief demonstration of each of the devices by writing the title of the source document. The participants were then asked to write their first names twice in order to obtain an understanding of how to operate the devices. During the practice sessions the participants were free to ask questions regarding the devices. During the Morse code sessions a chart of Morse code symbols and the source document were mounted to either side of the monitor at a height approximately equal to the screen. The participants were able to use the chart for reference throughout the study while typing from the source document. The participants used each device until their typing speed reached plateau.

The typing sessions were conducted in a quiet, well-lit computer lab on a college campus. Each typing trial was 20 minutes long, and participants were limited to no more than three trials in each session to avoid speed changes due to fatigue. Participants were allowed to perform no more that two sessions of three trials, separated by a minimum of one hour, in a single day.

Before each session, the input device to be used was set-up and ready to use, and a blank document in Microsoft Word 2000 was opened on the screen. The participants were given a different 500-word segment of the source document during each trial to replicate within a 20-minute period. The source document was placed level with the monitor on the side that was preferred by the participant.

At the beginning of each trial, the tester presented the source document, and told the participant to "type as quickly and as accurately as you can". The tester then would say, "Begin" and at this instruction a countdown timer (Cronus Olympian, single event timer) set for 20 minutes was started. After twenty-minutes of typing, the tester instructed the participant to "Stop," the tester deleted any partial words at the end of the trial, and the typing trial was saved to the computer's hard drive.

Data Analysis

After each session, the number of words typed within 20 minutes was determined using the Word Count feature of Microsoft Word. The words-per-minute typing rate was determined by dividing the total words produced by 20, the number of minutes in the trial. The words per minute change for consecutive sessions were compared to determine when a change of less than 7 percent was produced over 3 consecutive trials. One participant produced text at a rate that varied greater than 7 percent between sessions throughout the course of this study. This participant was unable to meet the 7 percent criteria between sessions to reach plateau, but was considered to have achieved plateau after having completed 9 trials with no upward change in performance.

Errors in typing were assessed by comparing, the typing produced by the participant with the source document using Microsoft Word's "Compare Documents" feature. The total number of differences between the source document and the participant's typing were counted (discarding the "difference" of an uncompleted sample) and recorded as the error count of the trial. In order to control for counting error, two testers reviewed each block of typed text until the same number of errors was reported from each tester.

The typing speed at plateau was computed as the average of the last three trials for each device. This moving-average was compared across devices to find the faster input method for each subject. Table-1, shown below, displays the average speed and accuracy for the last three trials for each of the participants in this study.

Table 1. Speed and accuracy of subjects using Morse code and Headpointer for text input.

Subject Number

Morse Code

Head-pointer

Speed

Accuracy

Speed

Accuracy

Naïve Morse code Users

221

5.7

96%

6.2

92%

426

2.2

96%

4.4

98%

507

4.9

95%

5.0

96%

738

4.1

100%

4.6

98%

Experienced Morse code Users

325

5.2

96%

4.7

89%

359

5.3

95%

4.4

95%

374

6.4

98%

4.3

93%

821

4.9

97%

3.5

96%

Figure 1. Typical Morse Naive Subject
(Click image for larger view)
Typical Morse Naive user reached plateau slightly above 6 words per minute with head pointer, slightly below 6 words per minute with Morse.

Results

All four participants who had no prior experience with either input method were able to type faster with the combination of head-pointing and on-screen keyboard than with Morse code. Figure-1 illustrates the typing speed of a typical participant with no experience on either input method. The non-experienced user reached plateau in fewer trials with the head pointer than with the Morse code device, and achieved a typing speed faster than that obtained with Morse code.

All four participants with Morse code experience reached a plateau at a greater rate with the Morse code device than with the head-pointer. Figure-2 illustrates the typing of a Morse experienced participant who reached a plateau in typing speed at a greater rate with the Morse code device than with head pointing. The participant reached plateau with the head pointer in fewer trials than the Morse code device but at a lower speed. However, the participant continued to demonstrate improvement in Morse code input upon reaching the standard for plateau.

Figure 2. Morse Experienced Subject (Click image for larger view)
Morse Experienced subject reaches plateau at about 5.5 words per minute with Morse, and about 4.25 words per minute with head pointer.

Accuracy of typing trials for all participants was visually compared in Table-1 for differences between the two input methods. There were no systematic differences in accuracy between experienced and non-experienced participants or between input samples produced by the two devices.

Discussion

Frequently, in a clinical setting, the decision as to which input device is best suited for the client is determined by a short-term assessment of speed and ease of use. Due to time constraints imposed by length of stay, the clinician is forced to make equipment recommendations without the benefit of long-term trials. Head pointing systems, given their low demands for training, offer fair performance with minimal training. The results of this study show that, over the short term, the client who has no experience with either input device is likely to demonstrate faster typing speeds using head-pointing than with Morse code. Hence, based on available information, the clinician is likely to suggest head pointing as the preferred input method.

Subject 507, a naïve Morse code user, illustrated a typical example of what is seen clinically. Figure-2 shows that head-pointer plateau was reached after five trials whereas 16 trials were performed before a Morse code plateau was reached, at a lower speed than was achieved with the head pointer. From this data, a clinician's decision will likely be in favor of the head-pointer.

The long-term benefit of Morse code is more clearly evident in the data collected from subjects 325, 359, 374, and 821, all experienced users of Morse code. These four subjects each performed significantly faster with the Morse code input both initially as well as in the long-term. This result indicates that choosing the headpointer based on a few trials may be a premature decision. It is likely that, with experience, the naïve users would demonstrate comparable results. It is important to point out that the experienced users were using an unfamiliar input method (sip-and-puff), and, because computer-based Morse code includes codes that are not used in Ham radio (such as shift and return keys), their initial speeds were somewhat below their nominal performance.

The long-term performance advantage of Morse code may derive from two aspects. First, Morse code offers the potential for becoming automatic, as illustrated by the reports of experienced users. Once the Morse code alphabet is learned, the amount of cognitive overhead significantly decreases, allowing more focus on the task to be performed rather than on the process of performance. While the position of letters on an on-screen keyboard also become over-learned, the process of guiding the mouse pointer to the desired screen location does not become as automatic. This is especially true considering the issues of pointer drift, which occurs with head-pointers due to the mouse acceleration curve . Hence, the on-screen keyboard user must consciously attend to the position of the pointer on the screen, slowing overall typing as compared with the automaticity of Morse code (Anson, 2001) .

A second potential performance advantage of Morse code is that it requires no visual feedback. Experienced Morse users attend to the "side-tones" of the Morse input system to hear the letters and words produced. While the on-screen keyboard user must constantly change visual focus between a source document and the keyboard, the Morse user can maintain visual attention on the source document at all times. This decreases the time spent hunting for the current position in a source document. This is similar to findings noted in the MacDonald et al study (McDonald et al., 1982) . In this study, the naïve participants needed to refer to a chart when producing the code because they did not know it from memory, as did the experienced users. This could have been a factor in explaining why the naïve users were not able to type as fast as experienced users of Morse code.

While this component of advantage would seem to apply only when working from copy, there is reason to believe that it also applies when composing text. Cognitive psychology hypothesizes the existence of two separate language centers: one for processing visual language, and a second for processing auditory language. The person who is using an on-screen keyboard processes both the keyboard on the screen and newly composed text visually, and must use only the "visuospatial sketchpad." Because Morse is processed as tones, it can occupy the "auditory scratch pad" or "phonological loop" (S. Reed, 2000) . This allows the text generated on the screen to remain in the visuospatial sketchpad and minimizes the mental exertion of composition.

Conclusion

When choosing an input method for a client, head pointing, in most cases, will allow for faster text input over the short-term because less learning is required. However, the client will not be using the selected input device only over the period of evaluation. For the client, this decision will affect performance over a period of many years. This study indicates that, given time, Morse code is likely to provide faster text generation than head-pointers. This conclusion is further strengthened by reports from the field, where Morse code users typically report faster typing speeds than head-pointer users.

This study compared Morse code input devices and head pointer input systems for speed and accuracy. The findings showed that, when judged over the short term of a clinical assessment, the combination of on-screen keyboard and head-pointer provides input rates that are faster than those achieved by Morse code, supporting the typical clinical behavior of selecting this more common input method. However, individuals who have experience with Morse code were able to type faster with Morse code than with the on-screen keyboard. Since individuals with disabilities will not be short-term users of the input method selected by clinicians, these results indicate that the solution justified by short-term use may not be the correct one.

Short-term clinical trials remain important in the selection process, since not all individuals with high-level disabilities are able to use both systems, and clinical testing is essential to uncover those limitations. For example, some individuals are able to position their head in space, but are not able to produce a discrete switch closure that is independent of head movement. While it is possible to select letters from an on-screen keyboard by "dwelling" on the letter for a period of time, this approach involves a substantial performance deficit. While Morse code can be used with head-switches, input speeds are substantially slower than with sip-and-puff input. In these cases, where good head position remains possible, on-screen keyboards might provide faster input over time. In other cases, the demands of head positioning might make an on-screen keyboard too fatiguing to use over time, and Morse would be the better long-term option. The results of this study apply only to those individuals who have the skills and endurance to use either system.

One weakness of this study is that the combination of head pointing and on-screen keyboard was not tested with experienced users. It is possible that, given time, there would be gains in performance for these individuals that are similar to those for Morse code. Prior research (Kanny & Anson, 1991) indicates that, over a few hours of training, head-pointer users achieve performance similar to that of able-bodied users of the mouse. Any performance gains, therefore, would be restricted to improved performance with the on-screen keyboard. Further research with long-term users of on-screen keyboards might give an indication of potential performance with this technology. However, a comparison of clinical reports from long-term users of head-pointing systems with reported speeds for Morse code users supports the findings of this study.

Further research might also explore the relative performance of these two input methods during composition tasks, where cognitive demands would influence performance. The added burden of composition might differentially change the rate of typing, with the more difficult input method being slowed further. It is also possible that the difficulty of using the input method might change the performance of composition.

In conclusion, while this study shows that the typical selection of head-pointing with an on-screen keyboard as an input method is justified by the data available through short-term clinical trials; it also shows that, over time, it may not be the optimal solution for a client. Over time, as Morse code becomes as automatic as touch-typing, it is likely to provide input rates that are substantially higher than those obtained by head-pointer users.

When both input methods are usable for a given client, the clinician should work with the client to encourage continuing the use of Morse until its performance can begin becoming automatic. Until that point, the client is likely to find it as cumbersome and annoying as hunt-and-peck typing. Only once it becomes as automatic as touch-typing, which experienced users indicate that it will, do the long-term benefits become evident.

Acknowledgements

We thank the participants of this study for generously providing their time: Dr. Joseph Cipriani, M.S., OTR/L; Robert N. Alder, M.S.; and the Wilkes-Barre HAM Radio Club. This study was performed in partial completion of requirements for a Master of Science degree of the second, third, fourth, and fifth authors at Misericordia University.

References

Americans with Disabilities Act of 1990(1990).

Angelo, J., Deterding, C., & Weisman, J. (1991). Comparing three head-pointing systems using a single participant design. Assistive Technology, 3 (1), 43-49.

Anson, D. K. (1997). Alternative Computer Access: A Guide to Selection . Philadelphia: F. A. Davis.

Anson, D. K. (2001). Assistive Technology. In L. W. Pedretti & M. B. Early (Eds.), Occupational Therapy: Practice skills for physical dysfunction (5th ed., pp. 257-275). St. Louis, MO: Mosby.

Anson, D. K., George, S., Galup, R., Shea, B., & Vetter, R. (2001). Efficiency of the Chubon versus the QWERTY keyboard. Assistive Technology, 13 (1), 40-45.

Beukelman, D. R., Yourkston, K. M., & Dowden, P. A. (1985). Communication augmenation: A casebook of clinical management . Austin, TX: ProED.

Chubon, R. A., & Hester, M. R. (1988). An enhanced standard computer keyboard system for single-finger and typing-stick typing. Rehabilitation Research and Development, 25 , 17.

Clement, M. (1961). Morce code method of communication for the severely handicapped cerebral palsied child. Cerebral Palsy Review, September-October , 15-16.

DeVries, R. C., Deitz, J., & Anson, D. K. (1998). A comparison of two computer access systems for functional text entry. American Journal of Occupational Therapy, 52 , 656-665.

Frame, C. (1994). Keynote speaker reviews history, predicts future of assistive technology. Retrieved February 26, 2003, from http://www.wiu.edu/users/mimacp/wiu/acct/C_Frame_Speech_at_ACTT_V.html

Jarus, T. (1994). Learing Morse code in rehabilitation: Visual, auditory, or combined method? British Journal of Occupational Therapy, 57 , 127-130.

Kanny, E. M., & Anson, D. K. (1991). A pilot study comparing mouse and mouse-emulating interface devices for graphic input. Assistive Technology, 3 (1), 50-58.

King, T. W. (2000). Modern Morse Code in Rehabilitation and Education: New Applications in Assistive Technology . Boston: Allyn and Bacon.

Krishef, C. H. (1991). Fundamental approaches to single subject design and analysis . Malabar, FL: Krieger.

McDonald, J. B., Schwejda, P., Marriner, N. A., Wilson, W. R., & Ross, A. M. (1982). Advantages of Morse code as a computer input for school aged children with physical disabilities. In Computers and the Handicapped. Ottawa: National Research Council of Canada.

Mergenhagen, P. (1997). Enabling Workers. American Demographics, 19 (7), 36-37.

Reed, K. L. (2001). Quick reference to Occupational Therapy (2nd ed.). Gaithersburg, MD: Aspen.

Reed, S. (2000). Short-term working memory. In M. Talfinger (Ed.), Cognition: Theory and Application (5th ed., pp. 104-106). Belmont, CA: Wadsworth/Thomson Learning.

HeadMaster Pro, Prentke Romich Company, 1022 Heyl Road , Wooster , OH 44691 , 1-800-262-1984, http://www.prentrom.com

Microsoft Word. Microsoft Corporation, One Microsoft Way , Redmond , WA 98052-6399 . USA

Dacri Card. The Darci Institute of Rehabilitation Engineering, 810 W. Shepard Lane , Farmington , Utah 84025 . Phone: 801-451-9191, Fax: 801-451-9393

Enabling Devices and Toys for Special Children, Inc. 385 Warburton Ave. Hastings-on-Hudson, NY 10706. Phone: (914) 478-0960

Madentec, Ltd., 9935 - 29A Avenue, Edmonton , Alberta Canada T6N 1A9 , Phone: (780) 450-8926, Fax: (780) 988-6182, Toll Free: (877) 623-3682, http://www.madentec.com.

Project Gutenberg (website), http://promo.net/pg/