Statistical disambiguation of multi-character keys applied to reduce motor requirements for augmentative and alternative communication

Abstract

The selection method employed in a communication aid is determined primarily by the ability of the user to interact accurately and rapidly with the device. The size of the targets (input switch, keys, symbols, etc.) and the number of targets combine to affect the overall success of the user. In the case of an orthographically based communication aid, a minimum of 27 keys or targets must be employed to allow unique direct selection of the letters of the alphabet and space. This approach is possibly compromised in the instance where a user's physical ability does not allow him/her to resolve the individual targets or extend over the range of targets. The effect of this is to introduce additional input errors, and/or reduce the input rate, and can result in the use of slower augmentative and alternative communication (AAC) methods such as encoding and scanning.This paper presents a new model for a communication system that reduces the number of keys to represent the alphabet from 27 to 11. The proposed keyboard employs nine keys, each representing a cluster of three letters, a tenth key dedicated to the space, and an eleventh used for error correction. The model uses fourth-order transitional probabilities to disambiguate among the characters of the three letter clusters. The model has been evaluated with several arrangements of letters against several texts. Measures are presented for the performance of the model in terms of the average keystrokes per letter and percentage correct disambiguation.The effectiveness of any aided augmentative and alternative communication (AAC) approach is largely dependent upon the interface between the user and the device. The ability of the user to make necessary motor input is a determining factor in the selection of a communicating system. That communication system, in turn, plays a significant role in determining the rate of communication. This is evident in the comparison between a system which allows direct selection of items (one keystroke/selection) requiring refined motoric ability and an alternative system in which items are selected by means of coded access (multiple keystrokes/selection) requiring relatively less refined motoric ability. While the language units (letters,, words, symbols, etc.) may be the same in both systems and the resulting message may take the same form, reduction in motoric requirements at the expense of additional keystrokes increases the time necessary to compose the message. If, however, the physical requirements of a communication aid can be reduced while not significantly increasing the number of keystrokes or access time, then the performance of the user/device combination need not be dramatically affected.This can be seen in the performance of users of 192 direct selection communication aids when the rate of output is limited by either key selection accuracy (resolution) or range of motion. Such interface difficulties exist in systems using manual, head position, and eye gaze selection techniques. While the direct selection of characters maint...

Probabilistic character disambiguation for reduced keyboards using small text samples

“…Three of these layouts were derived from previous research on reduced keyboards by other researchers Foulds, Soede, van Balkom, & Boves, 1987;Kamphuis & Soede, 1989;Levine et al, 1987). The "TOC" format seen in Figure 1(a) was developed by from an alphabetic layout by relocating the "t," "o," and "c" characters in order to minimize the disambiguation errors made by the system.…”

“…The quad-gram tables for the previous English and Dutch systems Foulds, Soede, van Balkom, & Boves, 1987;Kamphuis & Soede, 1989) were derived from large text corpora: 750,000 American English words from the Brown Corpus (Kucera & Francis, 1967) and 1 million Dutch words from the Eindhoven Corpus (Uit den Boogaart, 1975). The frequency of occurrence of any word or n-gram (i.e., the number of times that word or n-gram occurs) in such a text sample can be counted, and statistics can be gathered to indicate which words (or n-grams) occur most frequently in the sample and which ones occur less frequently.…”

Ambiguous keyboards for AAC

“…Minneman's work was quickly built upon, with a number of authors investigating the topic (Kondraske & Shennib 1986;Sh Levine et al 1987;Foulds 1987;Kreifeldt et al 1989; Sh Levine & Goodenough-Trepagnier 1990;Arnott 1992). Areas investigated by these authors include using syllable level disambiguation (i.e.…”

“…Gong & Tarasewich, 2005) although there is some investigation into character or prefix based disambiguation (I. Scott Letter distribution: Conventionally, letters are arranged alphabetically on a telephone keypad, this arrangement is purely conventional and is not optimal for using a disambiguation process. Foulds (1987) demonstrated some small changes in typing rate using alternative layouts on a standard phone keypad. Significantly larger changes between layouts are noted however as the number of keys is reduced (Venkatagiri, 1999).…”

Assistive Technology and Affective Mediation

“…These keyboards prove to be extremely useful, for example, when sending SMS (Short Message Service) communications on mobile phones, as well as helping users with special needs. The usual key distribution is the T9 keyboard 1 , however, there are studies that try to obtain better distributions (Arnott & Javed, 1992;Foulds, Soede, & Van Balkom, 1987;Gardeazabal, 2000;Lesher, Moulton, & Higginbotham, 1998a;Levine, Goodenough-Trepagnier, Getschow, & Minneman, 1987). The disambiguation that the reduced keyboards achieve can also be used as a prediction system.…”