KEY FACTORS IN THE SUCCESSFUL USE OF VOICE RECOGNITION TECHNOLOGY IN THE WORKPLACE

Dr. Tanya Goette, Assistant Professor
Georgia College & State University
Milledgeville, GA 31061
tgoette@solar.gac.peachnet.edu

Web Posted on: November 30, 1997

The purpose of this research was to understand why some people utilize adaptive technology (AT) successfully and others do not. AT can play a significant role in responding to the requirements for businesses as set forth in Title I of the Americans with Disabilities Act. This section requires that an employer provide a "reasonable accommodation" (Equal Opportunity Commission & the U.S. Department of Justice, 1991, p. I-9) for an employee with a disability. AT may be used to provide this accommodation.

The type of AT studied in this research was voice recognition technology (VRT). The purpose of VRT is to allow input without the use of a keyboard. VRT uses a microphone instead of a computer keyboard, thus enabling the individual to speak into the microphone instead of typing in commands.

A field study was used to examine the factors contributing to the successful use of VRT by individuals with disabilities. Interviews were conducted with 23 individuals who were successful and 15 who were unsuccessful to elicit their views on the use of VRT. As a result, it was possible to make informed comparisons between individuals with disabilities successfully using VRT and individuals with disabilities who were unsuccessful in their use of VRT.

VARIABLE MEASUREMENT

In this study, the dependent variable is success while the independent variables are categorized as expectations, innovation characteristics, and task-technology fit.

Four constructs constitute the independent variable of expectations. The expected benefits from using VRT, the expected success of using VRT, and the expected complexity of using VRT are important perceptions held by the adopter. The expectations that the adopter has about the expectations of others concerning the success of the adopter in using the voice recognition technology are also thought to influence the adoption outcome.

The perceived characteristics of relative advantage, compatibility, complexity, result demonstrability, visibility, and trialability are the characteristics of the innovation measured by the questionnaire.

The constructs comprising task-technology fit are the tasks of computer access, environmental control, word processing, and using VRT for all computer access. If VRT is used for environmental control, word processing, and all computer access, then there should be a suitable level of task-technology fit.

Use and user satisfaction as perceived by the user are utilized to form the dimensions of success for the purpose of this study. Use is divided into the level of use and achievements through use. The level of use is how often the VRT is employed while achievements through use is the percentage of tasks that are being accomplished through the application of VRT. Level of use was recorded in hours per week that the VRT is utilized as reported by the individual. Achievements through use was reported by the user in percentage of tasks accomplished of possible tasks performed by the VRT. User satisfaction was measured by the questionnaire.

RESULTS

In the sample interviewed, 25 individuals have quadriplegia due to intervertebral disk disorders, and two of these people are on respirators. Five others have quadriplegia due to multiple sclerosis, muscular dystrophy, or other nervous disorders. Seven individuals have motor skill problems due to cerebral palsy or muscular dystrophy. Additionally, five individuals have motor skill problems due to arthritis, tendinitis, or other joint or muscle problems. One individual has low vision, and two others have blindness with additional joint or muscle problems. The length of time the research subjects have had debilitating disabilities ranges from 2 to 48 years with the average length of time being 20.33 years.

Quantitative data analysis found four areas where successful and unsuccessful users were significantly different and two areas that were significant but in the opposite direction than was expected. Specifically, two hypotheses about expectations were significant but in the opposite direction (expected benefits and success) while expected complexity was found to be higher for successes. Support was found for two innovation characteristics hypotheses: perception of benefits and result demonstrability. Only one task- technology fit hypothesis was supported: that individuals successfully using VRT use VRT for all their computer access.

Discriminant analysis was used to determine which of the significant factors is the most important for distinguishing between successful and unsuccessful VRT users. There are three variables that are the most significant in differentiating between successful and unsuccessful use of VRT. Benefits of the innovation, which is a combination of relative advantage, compatibility, and ease of use, is the most important. This is followed by result demonstrability, and use of VRT for all computer access tasks.

An analysis of the statistical data indicated the differences between the successful and unsuccessful users of VRT, but why were the unsuccessful users unsuccessful? A cross-case analysis was conducted to organize the answers from each individual into groups (Patton, 1990). Four patterns emerged from the cross-case analysis of the interviews with the individuals who were unsuccessful in their use of VRT, and these patterns are discussed below.

PATTERNS FROM USERS

As demonstrated in research by Goodhue (1992), it is important that the technology fit the task it will be used to accomplish. Several individuals had difficulty in this area. Although VRT is a tool that can be used for a variety of tasks, some tasks can only be easily accomplished by certain types of VRT systems. For example, several individuals used software that allowed them to do drafting. If the individual used a VRT system that was specifically designed for drafting then the VRT could be used successfully. However, if the individual used a large vocabulary VRT system (ideal for word processing) for drafting, the individual grew frustrated with the process. Trying to use this type of VRT system for drafting is like trying to use Windows without a mouse--it can be done, but most users believe it is not worth the hassle.

On the other hand, some individuals had VRT systems that had a limited vocabulary. With these systems, a vocabulary must be set up in advance and made active in order to be used to accomplish a specific task. For example, one vocabulary is set up for use with a spreadsheet program while another vocabulary is set up for a database program. This type of VRT system works well for these limited vocabulary tasks. However, if the individual needs to do a lot of word processing, it soon becomes frustrating. To do word processing with this type of system, the individual must use the international alphabet to spell every word.

These examples demonstrate the importance of matching the task of the individual to the type of VRT system to be used. Individuals said that they believed part of this problem came from a lack of technological product knowledge on the part of themselves and the specialist they worked with to obtain the technology. Two solutions emerge to assist in eliminating problems with task-technology fit. One is to employ a technology specialist that oversees the task-technology fit of each purchase. A second, potentially better, solution is to allow individuals to try the product in their home for an extended period of time. This does require an initial training program for individuals who may ultimately decide not to adopt. However, it would probably be cheaper to conduct more initial training periods resulting in success stories from adopters than to purchase equipment that does not fit the task and ultimately is abandoned.

A second area that was cited as a reason for unsuccessful use of VRT by individuals was a lack of training. In the cases that seemed to have the best training, the individual was either trained by a consultant knowledgeable in training or by a person that concentrated solely on technology training. In the latter case, the person conducting the training was employed by the organization that purchased the equipment.

The amount of training received varied greatly. Twenty-two percent of the individuals received more than 20 hours of training while 26% received no training. However, two of the individuals with no training claimed they did not need any training because of their previous computer experience. Furthermore, the amount of training needed varies from individual to individual and is influenced by prior computer experience, prior knowledge of the software programs that the VRT will be used to access, and an understanding of how the VRT system works.

Individuals who are not computer literate should obtain more training than individuals who have substantial computer experience. A problem mentioned by organizations obtaining VRT systems for individuals with disabilities was the training time advertised by the system vendor. When the vendor stated that an individual only needed eight hours of training to successfully use the system, the agency purchasing the system may be allowed to pay for only eight hours of training for the individual. Vendors should modify their estimated training requirements to cover the difference in prior computer knowledge of their consumers thus allowing agencies flexibility in funding training.

A second training problem occurred in the workplace. At least three of the individuals who needed to use the system at work were not allowed time to get adjusted to the VRT system. These individuals may have had access to adequate training, but they did not have enough time to become competent in using the system. When the individual was given the VRT system, there was no reduction in work load or an extension of deadlines to allow for the extra time needed to learn to use the system. Therefore, the individual reverted to accomplishing the task using the previous method in order to complete it on time rather than taking the extra time required to complete it by using VRT.

A third problem with training was how it was accomplished. Usually, training occurred over a period of a day or two on a one-on-one basis. While the one-on-one basis works well, it may be better to stagger the training. If the individual learns a little about the system and is allowed to practice it over a few days or a few weeks, the individual may absorb more from the same amount of training. Also, if an alternative method of training were available, such as a videotape, the individual may learn more on his or her own. A few vendors offer videotapes, but these tapes are more of an advertisement than documentation. Many individuals with disabilities cannot use the VRT documentation manuals without extensive help; therefore, putting these manuals on audio cassette or videotape may aid the training process.

Several problems emerged that were related to the environment in which the VRT system was used. One problem was background noise. Some work environments are noisy, and this noise is picked up by the VRT system as a word to be input. In these situations, the individual has to constantly correct errors made by the background noise. Using the VRT for a trial period in the user's environment is one way of determining if background noise is a problem.

Another solution to cutting down on background noise is a change of microphones. Two types of microphones are available, desk microphones and headsets. A desk microphone picks up more background noise than a headset; however, the individual may need help to put on a headset. A headset microphone may have more accurate recognition because the individual is the same distance away from the headset microphone each time the VRT system is used. Therefore, increased accuracy and reduced background noise can result from using a headset microphone if aid is available to put on the headset.

A second environmental problem is the noise created by the individual talking to the computer. In some situations, other employees are distracted and cannot do their own work. In other situations, the information to be entered into the computer is of a private nature and should not be overheard. A simple solution to this problem is to put the individual using the VRT system in an office with a door that may be closed. This also reduces problems with background noise.

A final area of concern for individuals that were unsuccessful in the use of VRT is the limitations of the disability. VRT is not as fast as keyboard entry. Until it is, any individual that can type with one finger, one hand, or a typing stick faster than 25 words a minute will probably not be satisfied with VRT. Therefore, a technology specialist needs to be sure that other typing alternatives are tried before purchasing a VRT system. Sometimes the individual with the disability insists that VRT will do the job, only to discover that it is unacceptably slow. A trial period using the VRT would uncover this problem.

Another disability problem occurs when individuals have speech impediments. In order to work, VRT must have consistent input. VRT systems can understand someone with a serious speech impediment, even when other people cannot understand him or her, as long as the speech impediment is consistent. However, when the speech impediment is inconsistent the VRT cannot be accurate. Several individuals with cerebral palsy exhibited this type of problem; the VRT system would work successfully for some individuals and would not function properly for others. The easiest way to avoid this problem is to use the VRT for a trial period.

GUIDELINES FOR SUCCESS

Guidelines were formulated to be utilized when individuals with disabilities are adopting VRT. The application of these guidelines to other situations is limited.

The first guideline is that the expectations of the individual should be controlled to maximize the chances for success. Successful individuals expected benefits from using VRT, expected to be successful in using the VRT, and expected to have to work hard to understand and use the VRT. If individuals can see the VRT being used and see the results of using VRT, perhaps they will realize the potential of voice recognition technology.

To be successful, actual benefits from the use of VRT must be obtained by the individual with a disability. In order to achieve the potential benefits, the technology should be suited to perform the tasks it will be used to accomplish. It is also important that the amount of training correspond to the experience level of the individual adopting the voice recognition technology. In this way the individual can learn enough to be able to take advantage of all the functions available from the VRT system.

Qualitative findings suggest that trialability could be the most important contributor in avoiding the unsuccessful use of voice recognition technology. If individuals are allowed to try the VRT for an extended period of time in an actual setting, problems regarding the task-technology fit, disability limitations, and environmental concerns can be evaluated and possibly eliminated before adoption occurs.

ENDNOTES

1. This paper was adapted from Goette, 1995, Determining Factors in the Successful Use of Adaptive Technology by Individuals with Disabilities: An Exploratory Field Study, an unpublished doctoral dissertation, Georgia State University, Atlanta, GA.

REFERENCES

Equal Employment Opportunity Commission & the U. S. Department of Justice. (1991, October). Americans with disabilities act handbook (EEOC-BK-19). Washington, D. C.: Author.

Goodhue, D. L. (1992, June). User evaluations of MIS success: What are we really measuring? In J. F. Nunamaker & R. H. Sprague, Jr. (Eds.), Proceedings of the Hawaii International Conference on System Sciences: Vol. IV - Information Systems, (pp. 303-314). Los Alamitos, CA: IEEE computer Society Press.

Patton, M. Q. (1990). Qualitative evaluation and research methods. (2nd ed.). Newberry Park, CA: Sage Publications.