2001 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2001 Table of Contents


Word Cueing for Persons with Writing Difficulties: WordQ

Fraser Shein
Tom Nantais
Rose Nishiyama
Cynthia Tam
Paul Marshall
Bloorview MacMillan Centre
350 Rumsey Road, Toronto
Canada, M4G 1R8
E-mail: fshein@bloorviewmacmillan.on.ca

Introduction

Word prediction programs were originally designed for individuals with physical disabilities, to reduce the physical demands of typing. Relatively recently, these programs have been adopted by educators to provide assistance to students with learning problems who have difficulty with the mechanical aspects of text generation such as spelling and punctuation (MacArthur, 1996). Commercial word prediction products have gradually begun to include features specifically targeted for this application, such as integrated text-to-speech capabilities. This paper describes a new software package, called WordQ, which further integrates text generation assistance strategies with the software that people typically use for word processing on personal computers.

MacArthur et al. (1991) believe that the use of technology alone does not facilitate development of writing skills, instead, it should be used in combination with instructions on writing strategies. Technology is not used to correct deficits but rather to provide a compensatory approach that seeks to circumvent or ‘work around’ deficits while capitalizing on strengths (Raskind, 1994). Both word prediction and text-to-speech have been used for such compensation, allowing users to focus more on idea generation.

The success of word prediction for spelling assistance depends strongly upon the dictionary and the prediction algorithm. In addition, certain levels of visual and cognitive skills are demanded of the user. These demands are important considerations for individuals who may also have visual deficits, attention problems and other cognitive concerns. Proper configuration of word prediction software should help lessen these demands. Flexibility in presentation and interaction of the software is necessary to match the specific visual-cognitive needs and learning profile of the user.

Text-to-speech can also be a powerful writing tool for persons whose oral language skills are superior to their written language abilities. It enables these persons to use their general language sense to monitor their writing, and catch errors in grammar, spelling and punctuation that might otherwise go unrecognized (Hunt-Berg, et al., 1994; MacArthur, 1998). Visually highlighting words as the speech synthesizer reads them helps writers focus their attention on the words or sentences they have written. When text-to-speech is combined with word prediction, it can help users locate the target word in the prediction list. Lewis et al. (1998) found this combination to be more effective in reducing spelling errors than using word prediction alone.

Overall Design Considerations

The central philosophy behind the design of WordQ is that the product’s features and command functions should cater to the specific needs of writers with learning problems. As the design developed, it became clear that the software would need to go beyond simple word prediction and text-to-speech. In this larger view, word prediction can be seen as a particular example of a more general approach which we call word "cueing" – the process of supplying appropriate visual and auditory cues to assist the writer with text composition. Examples of word cueing follow in the discussion of WordQ’s features below. Other key design issues included the need for simple-yet-flexible design and presentation; minimal visual distractions; intuitive interaction; a customizable dictionary that matches a writer’s vocabulary; allowance for topical words; and integration of speech synthesis. A final issue, identified by MacArthur (1996) and Raskind (1998) was compatibility with standard applications. This design criterion ensures that individuals with learning difficulties have equal access to standard applications and it reduces the possible high cost and problems of portability and compatibility of information.

Specific Design Features

WordQ has two main windows that float over the user’s word processor (Figure 1). One of these windows displays word predictions as the user types and the other is a toolbar that provides access to WordQ’s commands and options. The word prediction features are designed for simplicity and flexibility. After the first one to three letters of a word are typed, the prediction window displays a list of several words, typically five, to complete typing the word. To help the user distinguish words, each word can be spoken in turn using text-to-speech. The user can control this auditory browsing with the keyboard or the mouse.

After completing a word, the WordQ uses knowledge of that word to predict the next word, including the correct word form. Such predictions are based on statistical information about how words are typically used together. As the user begins typing the next word, WordQ uses knowledge of the previous word to influence the words presented in the prediction list so that proper word forms are given priority. The knowledge base associated with the WordQ dictionary was generated by sampling a very large amount of text across the Internet at different writing levels. The vocabulary list for prediction is completely customizable -- WordQ can be tailored to an individual’s writing level and typical subject matter. A larger background dictionary spell checks and stores unrecognized words in a special review area for later acceptance or rejection. It adapts to the user’s writing style both in terms of word frequencies and common word combinations. As well, custom vocabularies can be created by importing text files containing typical text. Users and teachers can identify particular groups of words within a vocabulary as special "topics." When a topic is active (e.g., baseball), those words are given a higher weighting in the predictions.

Figure 1: WordQ supplies word predictions in a window that floats over the user’s word processor. Users can control WordQ’s behavior through the toolbar in the top right.

Text-to-speech is also used to enhance writing and editing. WordQ can echo letters, words, and sentences as the user types. Letter echo confirms typing a character. Word echo helps signal spelling errors; an incorrectly spelled word is spoken letter-by-letter. Word echo also confirms the selection of a predicted word. Sentence echo helps the user hear the word flow in a sentence to decide whether the right words and punctuation have been used. All of these echo features can be turned on or off from the WordQ toolbar.

A special text-reading mode is also provided to facilitate proofreading. With Microsoft Word, WordPad or Notepad, text can be highlighted and spoken word-by-word within that application (Figure 2). The user can also step individually between words and detect problems such as missing words. For other word processors (as well as non-editing applications like web browsers), selected text can be read back without the highlighting or navigation capabilities.

Figure 2: A special text reading mode is used for proofreading directly within the user’s word processing application.

Great efforts have been taken to keep WordQ simple and easy to use. It does not confront the user with excessive or confusing features. It is designed with those key features that make a real difference and with minimal and logical settings. The word list is visually simple so it does not distract the user and take attention away from the writing task. It can follow the text input cursor or it can be locked in any position. The following list describes the specific ease-of-use features incorporated into WordQ

Works under Windows 95/98/NT/2000
Functions transparently with any text editing program
Completes words using previous word knowledge and frequency-of-use
Predicts next word using previous word knowledge and frequency-of-use (option to turn off)
Predicts phrases
Predicts words at cursor location, including mid-word in existing text.
Allows undoing of word selections (Esc key)
Displays word list by most likely or alphabetic order
Displays word list vertically or horizontally
Displays user choice of up to 10 words (typically 5)
Allows user choice of word list font
Automatically sizes word list
Word list follows the text cursor or may be fixed
Allows word selection by either number line or number keypad, and/or mouse selection
Removes unwanted spaces and inserts correct spacing around punctuation following a predicted word
Supports international characters
Incorporates a user dictionary for individual writing styles (customizable; self-adapting; typically under 5,000 words)
Allows importing text files to customize user dictionary
Supports customizable topic word lists to focus prediction
Includes a background dictionary for spell checking (60,000 words US/Canada)
Holds novel words/spelling errors for later review
Allows novel words to be manually added while typing
Includes Microsoft text-to-speech engine
Allows word list to be spoken by browsing with arrow keys or mouse
Echos typing with speech (letters/words/sentences)
Reads text aloud within application (word-by-word highlighting and speaking words)

Acknowledgments

Financial support for WordQ's development was provided by the Ontario Rehabilitation Technology Consortium, supported by the Ontario Ministry of Health. A number of individuals have provided substantial technical assistance at various stages of the project: Shae Birch, Simon Frank, Fiona Kong, Colin Laine, Sherri Parkins, Pat Stoddart, Jeffrey Tucker and John Temprile.

References

Hunt-Berg, M., Rankin, J. L. and Beukelman, D. R. (1994). Ponder the possiblities: computer- supported writing for struggling writers. Learning Disabilities Research & Practice, pp.169-178

Lewis, R. B., Graves, A. W. Ashton, T. M. and Kieley, C. L. (1998). Word processing tools for students with learning disabilties: A comparison of strategies to increase text entry speed. Learning Disabilities Research and Practice, 13(2), 95-108.

MacArthur, C. A., Schwartz, S. S., and Graham, S. (1991). A model for writing instruction: integrating word processing and strategy instruction into a process approach to writing. Learning Disabilities Research & Practice, pp. 230-236.

MacArthur, C. A. (1996). Using technology to enhance the writing processes of students with learning disabilities. Journal of Learning Disabilities, 29(4), 344-354.

MacArthur, C. A. (1998). Assistive technology for writing. Perspectives, 24(2), 16-18.

Raskind, M. H. (1994). Learning disabilities in adult-hood. In P. J. Gerber & H. B. Reiff (Eds.), Assistive Technology and Adults with Learning Disabilities: A Rationale for Use. Reading, MA: Andover Medical Publisher.

Raskind, M. H. (1998). Assistive technology for individuals with learning disabilities: How far have we come? Perspectives, 24(2), 20-26.

WordQ is a trademark of Bloorview MacMillan Centre.


Go to previous article 
Go to next article 
Return to 2001 Table of Contents 
Return to Table of Proceedings


Reprinted with author(s) permission. Author(s) retain copyright.