Cynthia Tam, M.Sc., O.T. Reg. (Ont.), David Wells, Ph.D.

ABSTRACT:

This study evaluated the benefits of word prediction on rate and accuracy of typing when the prediction list was displayed with a PDA at the keyboard level. Eleven new users and ten experienced users participated in this study. With the new users, accuracy of typing was significantly higher with the PDA display. Experienced users typed significantly faster with the prediction on the PDA. Most users, new or experienced, preferred to have the prediction on the PDA. The differences in self-perceived performance or satisfaction after using WordQ with the PDA display were clinically significant in seven experienced users. Overall, this study showed that displaying word prediction at the keyboard level could improve rate and accuracy of typing.

KEYWORDS:

word prediction, location of prediction list, typing rate, typing accuracy, PDA

BACKGROUND

Word prediction computer software was initially designed to increase the rate of typing for individuals with physical disabilities [(1)]. However, studies have shown that reduction of keystrokes does not translate into an equal amount of improvement in rate [(2), (3)]. Researchers found that time required for visual-cognitive tasks such as scanning the prediction list and selecting the desired words offset some of the time benefits gained from keystroke savings [(3), (4)]. Adjusting parameters of the word prediction programs can lessen the visual-cognitive loads associated with its use. Displaying five words on the list in a vertical manner provide a balance between keystrokes saving and visual-cognitive loads [(5), (6)]. Changing the placement location of the prediction list on the computer monitor could also help to reduce visual search time [(7), (8)]. Previous study found that children with spina bifida preferred to have the prediction list to be placed at the lower border of the monitor so that their eyes did not have to travel very far to search for the desired word. Children in this previous study also had better accuracy of typing when using the prediction this way [(9)].

The study reported below evaluated the benefits of bringing the word prediction to the keyboard level. The specific objectives of the study were:

To assess the effect of the display location on rate and accuracy of typing
To assess participants’ preference for the location of the word prediction list
To assess the effect of the display location as perceived by the participants

METHODOLOGY

Participants:

Two groups of participants were recruited for this project. One group consisted of a random sample of 10 young people, aged between 11 to 14 who had used WordQ, a word prediction software, for one to two years [(10)]. A list of clients who started to use WordQ in the two years before the study period was compiled from the technology clinic’s database. Clients were then invited to participant in the study in a random order. The second group was a convenience sample recruited from new clients who were assessed at the technology clinic during the time of the study. This group consisted of 11 young people, aged between 10 to 18. Clients were invited to participate after they have been considered as appropriate users of WordQ by the assessing therapists. The second group of participants was introduced to WordQ for the first time. All participants typed using a standard keyboard with one to multiple fingers. They needed to look at the keyboard to locate the keys. Table 1 presents the demographic characteristics of the participants.

Table 1. Participants Demographics
Experienced Users				New Users
User	Age	Gender	Diagnosis	User	Age	Gender	Diagnosis
1	12	M	Cerebral Palsy	1	13	F	Medulloblastoma Brain Tumour
2	13	F	Cerebral Palsy	2	14	F	Spina Bifida
3	14	F	Developmental Delay	3	12	M	Ataxia
4	12	F	Cerebral Palsy	4	10	M	Developmental Coordination Disorder
5	14	M	Skeletal Dysplasia. Multiple Extostoses	5	12	M	Congenital CMV Infection
6	12	F	Sina Bifida	6	10	F	Acquired Brain Injuries
7	11	F	Cerebral Palsy	7	18	M	Cerebral Palsy
8	12	M	Beckwith - Wiedemann Syndrome	8	10	M	Ehlers Danlos Syndrome
9	11	F	Cerebral Palsy	9	17	M	Cerebral Palsy
10	11	F	Epilepsy	10	15	M	Cerebral Palsy
				11	11	F	Cerebral Palsy

Equipment:

Figure 1. Photo Showing the Set Up of the PDA Display (Click image for larger view)

A Personal Digital Assistant (PDA) that was connected to a computer was used to display the WordQ prediction list vertically. The PDA was embedded in a wrist rest (See attached graphic). Software utilities were developed for communication between the PDA driver and WordQ. The software changed the PDA display to be a prediction window that displayed five words vertically. It also allowed users to touch the words displayed on the PDA to select the desired word.

Procedures:

The study was approved by the organization’s ethics review board. All participants provided assents and their parents consented to have their children participate in the study.

For the experienced user group, an occupational therapist contacted the participant by phone and administered the Canadian Occupational Performance Measure (COPM) to measure participants’ perceived performance and satisfaction of performance on the story-writing task which was a common homework assignment for students in Ontario. A research assistant then visited the participants twice at their homes. At the first visit, a story was read to the participants, and they were asked to type using WordQ as they usually did for 5 minutes. If the participant had difficulties spelling the initial letters, assistance was provided. After that, the participants were introduced to using WordQ with the PDA. The 5-minute typing test was administered again after the participants had had a 10-minute practice of using WordQ with the PDA. The PDA was left with the participants to use for a total of 5 days. At the end of the 5 days, the research assistant visited the participants again. In the second visit, participant typed a story in the same manner as in the first visit, using WordQ on the PDA. A different story was used for each of the typing tests. The stories were matched in average letters per word to ensure that saving in typed letters using WordQ was similar in each story. Readability of all the stories was deemed to be at a Grade 5 level by the Microsoft Office Readability tool. The occupational therapist called the participant at the end of the second visit to obtain the post-measures on the COPM. Participants were also asked to indicate their preference for the location of the word prediction list. Their comments about having the prediction list displayed on the PDA were recorded.

For the new user group, data collection was completed as part of the participants’ computer access assessments. When the assessing occupational therapists decided together with the participants that the use of WordQ would be beneficial, the participants were asked to type using WordQ with the prediction list displayed at a location on the monitor as decided at the time of the assessment for 5 minutes. They would then type with the prediction list displayed on the PDA for another 5 minutes after a short break. The same stories and procedures used with the experienced users’ group were used with the new user group. At the end of the session, participants were asked to indicate their preference for the location of the word prediction list. Their comments about having the prediction list displayed on the PDA were recorded.

Data analysis:

Kump’s guidelines on calculation of typing rate and accuracy were used to score the typing samples [(11)]. Two occupational therapists scored 15 samples (36.6%) independently to establish the inter-rater reliability in the application of the scoring guidelines. The agreement was at 99.96%.

The effect of location of the prediction list on rate and accuracy, and the changes in the COPM Performance and Satisfaction Scores were analyzed with paired t-test using SPSS. In addition, the COPM scores were analyzed descriptively because a change of 2 in COPM Performance and Satisfaction scores was necessary for the result to be considered clinically significant [(12)]. Participants’ preference for location of word prediction list and their feeling about having the display list on the PDA were analyzed descriptively.

RESULTS

Data of two experienced users’ typing rates and accuracy were lost due to technical difficulties, therefore the data used for comparison of rates and accuracy from the user group were from 8 users, but the COPM data of all the users were used in the analysis.

Descriptive statistics on rates and accuracy of typing in both groups are presented in Tables 2 and 3. Comparing the rates of typing of the experienced users, significant difference can be seen between the typing rate with the display on the monitor and the rate at the second visit with the PDA display (p=0.04). Marginally significant difference is found in the typing rate between the first and second visit with the use of the PDA display (p=0.058). No statistical difference is detected in accuracy of typing with the display in the two locations. Statistically significant differences are found in the change of Performance (p=0.04) and Satisfaction (p=0.048) scores. The COPM performance and satisfaction change scores are clinically significant in 7 different users (See Table 4). Comparing the results of the new users, significant difference is seen in the accuracy (P=0.037) of typing but not in the rate of typing. For both groups of participants there was one person who preferred to have WordQ on the monitor. Others preferred to use the prediction list on the PDA.

**Table 2. Typing speed and accuracy of experienced users**
	Means	Standard Deviation
Typing rate with display on monitor	9.8 wpm	6.96
Typing rate with display on PDA (1st visit)	10.03 wpm	6.97
Typing rate with display on PDA (2nd visit)	11.33 wpm	7.1
Typing accuracy with display on monitor	93.67%	10.15
Typing accuracy with display on PDA (1st visit)	95.84%	5.04
Typing accuracy with display on PDA (2nd visit)	99.49%	0.83

**Table 3. Typing speed and accuracy of new users**
	Means	Standard Deviation
Typing rate with display on monitor	9.22 wpm	5.14
Typing rate with display on PDA	8.92 wpm	5.8
Typing accuracy with display on monitor	97%	3.02
Typing accuracy with display on PDA	99%	0.76

**Table 4. COPM scores of experienced users**
Users	First	Second	Performance Change Score	First	Second	Satisfaction Change Score
Users	Performance Score	Performance Score	Performance Change Score	Satisfaction Score	Satisfaction Score	Satisfaction Change Score
1	7	8	1	9	10	1
2	7	9	2*	8	8	0
3	5	8	3*	5	8	3*
4	6	5	-1	8	5	-3
5	6	10	4*	6	10	4*
6	7	8	1	6	8	2*
7	8	9	1	8	9	1
8	7	8	1	5	7	2*
9	7	9	2*	8	9	1
10	6	9	3*	6	10	4*
*clinically significant

DISCUSSION

The results of this study confirm the results of previous study that the location of the prediction list can affect typing rate and/or accuracy of word prediction users [(9)]. Many children with physical disabilities type with one or multiple fingers on a standard keyboard. They have to watch their fingers’ movements. Switching their eye gaze between the keyboard and the monitor takes a significant amount of time. Bringing the word prediction to the keyboard may have reduced the need of moving the eye gaze away from the keyboard, thus offering better support for the use of word prediction by children. However, the small sample size is this study needs to be noted as a limitation. Future studies with larger samples of individuals with various diagnoses and ages are required to confirm the findings of this study.

The use of a PDA as a display offered an additional benefit that has not been explored in this study. Many children with physical disabilities have difficulties with organizational skills and will benefit from using a PDA as an organizational tool. Our design allows the PDA to be disconnected easily and be used as an independent device, but further software development will be necessary to facilitate switching of the mode of the PDA between functioning as a prediction window and an organization tool.

REFERENCES

Heinisch, B., & Hecht, J. (1992). A comparison of six programs: Word prediction software. TAM Newsletter, 8(3), 4-8.
Goodenough-Trepagnier, C., & Rosen, M.J. (1988). Predictive assessment for communication aid prescription: Motor-determined maximum communication rate. The Vocally Impaired: Clinical Practice and Research. (pp. 167-185). Philadelphia: Grune and Stratton.
Vanderheiden, G.C., & Kelso, D.P. (1987). Comparative analysis of fixed-vocabulary communication acceleration techniques. Augmentative and Alternative Communication, 3, 196-206.
Horstmann, H., & Levine, S.P. (1991). The effectiveness of word prediction. Proceeding of RESNA 14th Annual Conference. Washington, DC: RESNA Press.
Swiffin, A.L., Arnott, J.L., Pickering, J.A., & Newell, A.F. (1987). Adaptive and predictive techniques in a communication prosthesis. Augmentative and Alternative Communication, 3, 181-191.
Venkatagiri, H.S. (1994). Effect of window size on rate of communication in a lexical prediction AAC system. Augmentative and Alternative Communication, 10, 105-112.
Klund, J., & Novak, M. (1995). If word prediction can help, which program do you choose? Paper presented at RESNA 18th Annual conference at Vancouver, British Columbia.
Newell, A.F., Arnott, J.L., Booth, L., Beattie, W., Brophy, B., & Ricketts, I.W. (1992). Effect of the "PAL" word prediction system on the quality and quantity of text generation. Augmentative and Alternative Communication, 8, 304-311.
Tam, C., Reid, D., O'Keefe, B., & Naumann, S. (2002). Effects of word prediction and location of word prediction list on text entry with children with spina bifida and hydrocephalus. Augmentative and Alternative Communication, 18, 147-162.
WordQ version 2.2 [Computer Software]. (2006). Toronto, ON, Canada: Quillsoft Ltd.
Kump, A. (1992). Direction for scoring typing tests taken either on a typewriter or a computer [Data file]. Rockville, MD: U.S. Department of Education, Office of Educational Research and Improvement. (ERIC Document Reproduction Service No. ED 356269).
Law, M., Baptiste, S., Carswell, A., McColl, M., Polatajko, H., & Pollock, N. (1998). Canadian Occupational Performance Measure. (3rd ed.) Ottawa, ON: CAOT Publications ACE.

Benefits of Displaying Word Prediction Window at the Keyboard Level