Web Posted on: February 16, 1998

IMPROVING THE INTERACTION RATES OF USERS OF ASSISTIVE COMMUNICATION DEVICES

Donald B. Egolf, Ph.D.
Department of Communication
1117 Cathedral of Learning
University of Pittsburgh
Pittsburgh, PA 15260
Voice Message: (412)624-6763
FAX (412)624-1878
Internet: ratchet@vms.cis.pitt.edu

A critical problem faced by individuals who use assistive communication devices is the reduced rate of interaction (Light, et al., 1990: Foulds, 1987; Yoder & Kratt, 1983). The rate problem can be addressed at the level of input, device coding, and output. The focus of the present study was on coding. Specifically, two investigations were conducted. In Study 1 a code was developed that was designed to increase the interaction rate of a device user who wanted to send both Quicktalk and Exacttalk messages (see Vanderheiden & Lloyd, 1986): and, in Study 2, an investigation was conducted to determine which of four possible coding schemes could be recalled most efficaciously.

In Study 1, three literate, cerebral-palsied, teen-age students participated. Each student had a sophisticated assistive communication device which was coded with Quicktalk utterances and a few fragments for Exacttalk communication. A 50-minute conversation was held with each student and the conversation was then transcribed. Using as a comparison index for interactional gains was the number of keystrokes needed if users just "typed and machine-talked" all messages. It was found that users saved respectively 17, 21, and 24 keystrokes (all keystroke numbers have been rounded to a whole number) with their current codes over the comparison index. In short, what would have taken them 100 keystrokes to 'type and talk' now only took 83,79, and 76 keystrokes. Could even further gains be made? To make this determination the Quicktalk/Exacttalk code (QT/ET code) was developed. The code was designed to facilitate the fast production of Quicktalk utterances (ritualistic utterances) and Exacttalk utterances (utterances that are novel). A formal analysis of the students' transcripts showed that the code would be even more efficient than the students' current codes. But could the students learn the new code and would they want to? They agreed to try, and were able to learn (with the aid of within-view reference cards) the new code within five, 50 minute sessions. A 90% criterion was used for success, i.e., the code was learned when the students used the code 90% of the time in which it was indeed usable. After learning the new code a second spontaneous conversation was held with each student. Analysis of that conversation showed that by using the QT/ET code the students had increased their interaction rates over that which would have been achieved had they used their current codes. The average savings was 8 keystrokes (respectively 11, 7, and 6). The results of Study 1 suggest that continued refinements, those that promote increased interaction rates, of user codes can be made. In addition, because many device users are motorically challenged, a single keystroke saving can have a dramatic impact on interaction speed over the course of a conversation. The present study is compatible with studies having similar goals (e.g., Higginbotham, 1992; Szeto, et al., 1993; Venkatagiri, 1993).

Study 2 investigated which of four code types was most efficacious in terms of recalling codes that triggered Quicktalk messages from users' devices: (a) icon codes created by users, (b) icon codes created by a professional, (c) letter codes created by users, and (d) letter codes created by a professional. Fifty nine undergraduates participated and were randomly placed in one of four code groups. Participants had three weekly learning and recall-testing sessions. The reason for multiple sessions was to test for improvement over time. Participants studied their codes for thirty minutes before each testing session. Testers would read each of 100 common discourse utterances, whose codes the participants were learning, and then ask the participants to give the code that would recall the utterance from a device's memory. The data were subjected to a three-way ANOVA with repeated measures. As expected, 'session' had a significant main effect upon the test scores (F(2,110)=143.43, p<0.0001). Students' scores improved across sessions. 'Type of code (letter vs icon)' had no significant (F(1,55)=3.03, p>0.05) main effect upon the recall even though there was a tendency for the letter-coded group to score higher than the icon-coded across sessions. There was a significant main effect for 'self coding vs professional coding' (F(1,55)=5.04, p<0.01). Those who encoded messages by themselves consistently showed higher recall scores across sessions than those whose codes were preselected by others.

The fact that self-coders (both letter and icon) performed significantly better than subjects who used pre-coded items agrees with the intuitions of many clinicians and teachers who believe that users should be involved in the selection of symbols for retrieval codes. A documented example of user participation in symbol selection is found in Falk's (1988) report on a nonspeaking, augmentative communication device using student. Falk notes that "the student's own input was invaluable in deciding which lexical items to assign to which symbols." Self-coders, it seems, bring to the situation their connotative meanings which are individual and which cannot be anticipated by a pre-coder. Whether or not self-coding alone remains a superior strategy as cognitive loads are progressively increased is another question.

The present study was nomothetic in nature and it stands in contrast to the ideographic or case study approach or one-on-one tutorial approach. On the surface it may seem that studies examining group data, particularly if that data were gathered on normal subjects, may have little relevance for the clinician or teacher working with an individual nonspeaker. Nonetheless, nomothetic studies can provide the confirming support for ideas that may arise in clinical or classroom endeavors, and may provide a repertoire from which clinical and educational strategies can be selected.

Falk, J. P. (1988) TouchTalker: A Case Study. "Communication Outlook," 102, 1-11.

Higginbotham, J. (1992) Evaluation of Keystroke Savings across Five Assistive Communication Technologies. "Augmentative and Alternative Communication," 8, 258-272.

Light, J., Lindsay, P., Siegel, L., & Parness, P. (1990) The Effects of Message Encoding Techniques on Recall by Literate Adults Using AAC Systems. "Augmentative and Alternative Communication," 6, 184-197.

Szeto, A., Allen, E., & Littrell, M. (1993) Comparison of Speed and Accuracy for Selected Electronic Communication Devices and Input Methods. "Augmentative and Alternative Communication," 9, 229-241.

Vanderheiden, G., & Lloyd, L. (1986) Communication Systems and their Components. In S. W. Blackstone (Ed.) "Augmentative Communication: An Introduction," Rockville, MD: American Speech-Language & Hearing Association.

Venkatagiri, H. (1993) Efficiency of Lexical Prediction as a Communication Acceleration Technique. "Augmentative and Alternative Communication," 9, 161-167.