2000 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2000 Table of Contents

Speech recognition in schools: An update from the field

Bob Follansbee
Education Development Center
Newton, MA

Susan R. McCloskey-Dale
Instructional Support System of PA
Hummelstown, PA

The use of speech recognition in schools represents a holy grail of assistive technology practice for many. Parents, students, and even some teachers have clamored for it, but successful use has eluded many as various factors stand in the way of implementation. This presentation will discuss issues required for successful implementation of speech recognition in schools and present case studies of two students to illustrate the points.

The presenters represent two different organizations: Speaking to Write and the Instructional Support System of the Pennsylvania Department of Education. The Speaking to Write project is a joint venture of the Education Development Center and the Computer Learning Program at Children’s Hospital, Boston. This project has been working with students, parents, teachers, and trainers in speech recognition as part of a federally-funded project to develop materials that support the use of speech recognition in schools. As part of its work, the Speaking to Write project has maintained a listserv for the past two years that has been a very successful source of online collaboration and information regarding the use of speech recognition in schools.

The Technology service of the Instructional Support System of Pennsylvania has been supporting the implementation of speech recognition among students for several years. They have developed implementation strategies and assistive technology implementation standards against which successful use of speech recognition can be measured.

Speech recognition presently comes in two styles: discrete speech and continuous speech. The older technology, discrete speech recognition, operates by requiring that the user speak one - word - at - a - time. A newer technology, continuous speech recognition, allows the user to dictate by speaking at a more or less normal rate of speech. Both have their advantages and disadvantages for individuals with difficulties in writing, although continuous speech recognition is the only one of these two that is currently receiving any attention from the marketplace or the speech recognition companies. In fact, discrete speech recognition is slowly being phased out and new operating systems will not support the older programs any longer.

However, conversations with listserv members has further confirmed a growing impression of project personnel that discrete speech recognition, despite its slow speed and poor market success, is a much better tool for some students with disabilities, for a variety of reasons. Typically, one would consider using discrete speech products now if there was some extenuating circumstance that limits one’s ability to use continuous speech, such as significant learning disabilities, motor speech difficulties, and so forth.

In our experiences with students, when presented with both options, many potential users with learning disabilities and other disabilities with similar profiles prefer the operation of discrete speech rather than continuous because they find the pacing and operational style provided by discrete speech better matches their text production style, at least at that point in their development. The slower pace better matches their own style or pace in creating written, the manner of correction makes it easier for them to find and make corrections, and so forth. At the same time, some students automatically gravitate toward continuous speech and manage very well with it, sometimes despite certain "mismatches" in their apparent abilities.

Successful use of speech recognition by students as an independent text input strategy requires consideration of those student characteristics that have been identified as important background variables, and adequate training. Additionally, successful use in schools also assumes some level of systemic support and accommodation. The student characteristics to consider include: general cognitive level, oral-motor speech abilities, reading and spelling abilities, oral language formulation and flexibility, self-monitoring ability, perseverance in the face of frustration, and a sense of the value of writing. Training in the use of speech recognition software is also critical, as one must learn not only how to operate the software itself, but how exactly to dictate to produce the best results. The process of learning to dictate by voice can be likened to the process of learning to type in terms of length of time for achieving a high level of independence.

The systemic support required of school for developing a climate for successful use of speech recognition would include: identification of resource people in terms of technical support for computer use and in terms of familiarity with the speech recognition software to assist the student when needed; commitment to adequate staff release time for training; identification of one or more places in the school where the student can use speech recognition satisfactorily; and provision of adequate instructional support for the student to "catch up" on lost writing opportunities.

We propose to provide examples of students using speech recognition in schools without unusual conditions or extraordinary efforts. The primary requirement for success, basic to many kinds of assistive technology, is that school staff, parents, and student all saw the benefits of speech recognition for the student in question and determined to make it work. This requires more training (for teacher and student) and integration planning time initially than many technologies, but this initial expenditure of resources drops with subsequent users. Obviously, success depends on systemic support from the school, adequate resources, an openness to technology-based solutions for struggling writers, and a climate that values writing.

In our view, speech recognition is simply another tool for writing. It should be seen as part of a continuum of writing technologies strategies that stretches from pencil to input by voice. In this view, speech recognition certainly lies at the most sophisticated end of this spectrum, and in most cases should not be considered until a variety of other less complex strategies have been tried. However, there are times when students who have struggled with more traditional writing methods pick up speech recognition very quickly and easily, and in these cases, there is little reason to insist that they go through intermediate steps. By the same token, speech recognition does not need to be one’s only method of text production if the student can use other strategies at other times.

Successful use of speech recognition technology is not simply a matter of producing more text more quickly. The standards based on words per minute that might be of interest to professionals who need rapid entry of highly repetitive text (e.g., medical records in radiology) are less relevant to students who are simply trying to find a better method of creating written work. Success for struggling writers is more likely measured in better performance on a variety of writing tasks in school, slowly increasing independence in producing written work, and more positive perceptions of themselves as writers and students.

Go to previous article 
Go to next article 
Return to 2000 Table of Contents 
Return to Table of Proceedings

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