WordPrediction

d

Use of Co:Writer can help build students' self-monitoring skills. Co:Writer (Don Johnston)

Word prediction also supports writers with automatic production difficulties. Drawing on a stored set of words, programs with word prediction generate suggested options for completing words during text entry. When the first letter is typed, the program presents a list of possible words beginning with that letter, based on both spelling and syntax. As each additional letter is typed, the list is refined. When the intended word appears in the list, the user selects it, by clicking on it or typing its number, and the program inserts the word into the document. Mechanisms for presenting and inserting words range from a scrolling word list in a separate window (My Words) to a pop up of frequently used words or phrases that can be inserted by pressing "return" (Microsoft Word). Co:Writer functions as a separate text-entry window with any word processor. When a sentence is completed it is automatically placed into the word processor document.

Co:Writer also provides text-to-speech support and uses a form of artificial intelligence to build the frequency lists for prediction. That is, words the writer actually types most frequently become the first on the list in the prediction window. This can be a plus or a minus, as students with writing difficulties often misspell a word the same way multiple times. Thus an incorrectly spelled word could top the frequency list. Fortunately, it is possible to clear and edit the list, an exercise that could be useful in helping build students' self-monitoring skills if done collaboratively.

Word prediction lowers the number of keystrokes, and supports children with severe spelling problems, as frequently only the first few letters of a word need be typed (MacArthur, 1998). Studies are also suggesting that word prediction may actually improve writing and spelling accuracy for students with learning disabilities. In a study on the effects of word processing with word prediction on the writing accuracy of handwritten response journals among a small sample of students with learning disabilities, MacArthur (1998) found that four out of five students improved their writing legibility and spelling accuracy. Lewis, Grave, Ashton, and Kieley (1998) compared LD and control groups using a variety of text entry methods including word processors with word prediction. Handwriting was the fastest overall means of text entry with word processing/word prediction next. All technology groups improved writing and spelling accuracy, and the gap between writers with and without learning disabilities narrowed on the dimension of text accuracy.

These findings suggest that word prediction has significant promise as an important writing scaffold for students with learning disabilities, and may even improve their routine skills. Because it interrupts the flow of typing, and therefore can be frustrating for students wishing to keep up with the flow of their ideas, it may be helpful to teach word prediction as a skill in its own right. Research is needed to clarify whether word prediction is most effectively taught when students are learning keyboarding or after keyboarding skills are automatized.