The advent of the microcomputer has provided the severely handicapped with the means to create text. Instead of using a keyboard, the disabled typist is able to scan and select linguistic items with an appropriate input switch. The resulting communication rate is, however, prohibitively slow for writing and impractical for conversation. A variety of techniques is used to improve this rate and range from static letter matrices to more sophisticated methods in which words and phrases are anticipated. Although many anticipatory methods claim to be linguistically based, most, if not all, depend solely on letter and word frequency statistics. A series of phonological rules can be used to anticipate the letter structure of most English words. This linguistically based system reflects a degree of "intelligence" not present in other anticipatory writing systems. To evaluate and compare the new system with several existing techniques in practice, a programmable evaluation system has been developed on an IBM-compatible personal computer using the Artificial Intelligence language, LISP. Different communication strategies are transcribed into rulebases which serve as input to the software. The core program then executes the particular system under consideration. Input text can be processed in either manual or simulation mode and an evaluation report is generated when the session ends. The characteristics of efficient communication systems are introduced as a basis for this dissertation, after which the development and application of a linguistic anticipatory writing system is described. The design of the evaluation software is documented and the successful implementation of the various communication systems is discussed.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uct/oai:localhost:11427/26608 |
| Date | January 1989 |
| Creators | Waller, Annalu |
| Contributors | Boonzaier, David |
| Publisher | University of Cape Town, Faculty of Health Sciences, Department of Human Biology |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Master Thesis, Masters, MSc (Med) |
| Format | application/pdf |
Page generated in 0.0018 seconds