The voice as a learning technology: a review.

Abstract

Voice recognition technologies (VRT) are learning technologies that enhance students' composing processes. They encourage instructors to ask questions concerning what benefits oral composing can bring. This essay reviews several studies (specifically those conducted by BECTA, "Speaking to Write," and Scotland's CALL Centre) that have explored students' interaction with voice recognition programs to produce texts. Drawing from these studies, which focus on the writing of younger writers, it also delves into what future areas we should explore about oral composing technologies in the college classroom.

Introduction

Writing teachers talk about "voice" in scholarship, but they employ the term as a metaphor for the persona that comes through in students' texts that are, generally, produced in silence. For example, Donald Murray writes that instructors should "listen carefully for those words that may reveal a truth, that may reveal a voice" (5) and that they must respect students for those potential voices. However, in neglecting truly oral composition as a form of "voice," instructors overlook an area of the writing process that can benefit students.

Voice recognition technologies, rather than simply a means of delivering text to screen orally, are learning technologies that enhance students' composing processes. Several researchers (some of the most notable examples are Sheryl Lee Day (education); Michael Chamberlain (education); John Reece and Geoff Cumming (composition); Carl Berieter and Marlene Scardamalia (education); Beatrice Bourdin and Michel Fayol (psychology); and Susan De La Paz and Steve Graham (education and learning disabilities)--have weighed in on questions surrounding the use of technology and dictation to make composing easier. Beyond these sources, it is crucial to look at the work done with student writers, as doing so offers a glimpse into how C-MOC (the term I will use to refer to computer-mediated oral composing) works in practice.

Projects completed by three international groups of researchers (In Britain, The United States, and Scotland)--have been conducted concerning the effects of speech technology on pre-college students' writing. The results of these studies, as well as the training materials generated from them, are groundbreaking for all instructors who integrate speaking/writing into the classroom. Reviewing the methodologies, findings, and further questions raised by these projects helps consolidate the current literature. Additionally, such a review illustrates how these studies' observations are essential in creating new pedagogies of multi-modal writing instruction.

Overview of Studies

Materials and findings from educational projects offer a perspective on oral composing that takes into account the experiences of student writers. Corporate training materials for speech software programs, such as Dragon Naturally Speaking, do address some aspects of the composing process (usually telling writers to do such things as speak as if conversing with a friend to increase accuracy). But they also tend to address composing in a more prescriptive manner, one not founded by knowledge of the complexities of the writing process. Three projects are discussed here: the multimedia and web-based training created by Scotland's Communication Aids for Language and Learning Centre (CALL), data from the British Educational Communications and Technology Agency (BECTA) Project (which focused on C-MOC for secondary students with disabilities), and the "Speaking to Write Project", carried out from 19972000. Each project discusses the process of C-MOC either through the lens of students' personal experience or through educational research. All three were long term inquiries into the medium's impact on younger students' writing processes. The data collected from the projects is relevant not just for the grade levels represented, but also for all students who may reap similar benefits from oral composing. Thus, the projects provide instructors in higher education with pressing questions about how oral composing can work in higher education.

The CALL Centre

The Scotland-based CALL Centre focused on voice recognition technology (both Dragon Naturally Speaking and IBM's ViaVoice) as a writing tool for elementary students with learning disabilities. The Centre's web-based materials include video clips of children working with CMOC and discussing the frustrations and benefits associated with the medium. Although the researchers comment that it is exaggeration to paint speech to text programs as a '"panacea' for people with dyslexia and other writing difficulties," they explain that many students have used them successfully for essays and exams (1). Methods of gathering information in this project involved student participation in the development of the results. CALL researchers included brief taped interviews with young participants who reflected on their experiences (citing both advantages and disadvantages) with creating text orally. One young boy cites the ability to produce more text faster as an advantage of oral composing, while another student discusses her frustrations with the errors in recognition. The Centre's many training materials for Dragon Naturally Speaking focus on agency and medium, providing suggestions for a delivery style when writing by voice:

   Speak the phrase or sentence clearly with a comma or full stop.
   Use a 'dictating voice', speaking slightly more slowly and carefully
   than you would for conversation. Try to speak like Eddie Mair, the
   newscaster--you can speak with an accent, but you must be clear and
   distinct. Avoid running words together, or letting your voice tail
   off at the ends of sentences. (32)

Though instructors may not want to present college composition students with a prescriptive style for creating text orally, the Centre's instructions highlight the need to address issues of voice and diction with our students. Their methods of using participant interviews brings the voice of the writer into the data and offers valuable first hand accounts of composing with technology. Significant results from the study relate to students' literacy and writing ability. Although a minority of students chose not to continue using VRT because of, among other problems, its errors, those who decided to continue composing orally benefited in several ways. They displayed increased oral skills and reading ability, greater writing quantity and quality, and a greater motivation to write and work independently (23). These results support the assertion that VRT is, indeed, a learning technology.

BECTA

The British Educational Communications and Technology Agency completed a two-year study (April 1998-March 2000) focusing on speech recognition's benefits for students. The Dfee/BECTA SEN Speech Recognition Project focused on CMOC's uses for secondary students with disabilities. The researchers' primary question was simple: "Do the systems really work and if they do, how can they work most effectively for pupils with special educational needs?" In order to carry out the project, the team was given 100,000 pounds by the Department for Education and Employment, which was used over a period of two years. The agency identified 12 centers for implementing C-MOC and made sure each was provided with funding. Although previous studies had avoided discussing VRT's effectiveness with different dialects, BECTA sought to have "a wide geographical spread in order to compare the systems with different accents." They looked for ranges in ability and age, a balance between experienced and new users (3). Despite technical difficulties, they noted that students using VRT were better able to "think through their ideas" and that controlling the PC through speech recognition gave them "the ability and time to structure their thought" (5). The researchers state that "there is no doubt that the software works and can liberate many learners by allowing them to produce text in greater quantity and quality than by any other method" ("What are Speech Recognition Systems?"). Their findings point to some promising benefits for writers with disabilities. One is that C-MOC allows visually impaired students more independence in their composing (6). And their conclusion about the greater quantity of text produced indicates the technology's potential as a powerful heuristic for students suffering from writer's block.

The researchers found that "pupils appeared more willing to experiment with content and layout, producing more elaborate and extended pieces of writing with a wider range of vocabulary and sentence structure." And in discussing C-MOC's affects on students' literacy, they write that "because the students needed to develop dictation skills, they became more able to bullet-point and draft out ideas more quickly." Interestingly, the examination boards forbade the use of VRT in exams, stipulating that the technology is reserved for extraordinary circumstances in Scotland (5-7). But when the benefits for pre-writing and formatting are so clear, it seems that such programs should not be reserved for use in special circumstances. Also in relation to literacy, the researchers comment that C-MOC encouraged students to use punctuation more correctly. They also observed that "by the end of each session there was a noticeable improvement in articulation and voice projection (6)." Their findings suggest that CMOC can encourage students to be not only more savvy users of punctuation, but also more attuned to the way their text sounds as they create it (thus giving the process an element of performance).

"Speaking To Write"

The "Speaking to Write" Project, which lasted from 1997-2000, and was funded by a grant from the National Institute Disability Research and Rehabilitation), was begun by Dr.'s Bob Follansbee, Patricia Corley, and Lucy Lorin. They studied the benefits of discrete VRT (a program that recognizes only one word at a time rather than strings of flowing sentences) for middle and secondary students with disabilities and produced training materials for instructors integrating C-MOC into their classrooms. The study produced a print manual and an interactive cd-rom designed to train younger students to use Dragon Naturally Speaking. Additionally, the project resulted in the "spk2wrt" listserv, through which teachers and parents of students with disabilities and VRT users discuss training/writing issues. The manual is "a hybrid of theoretical information and training tools" (10). Examination of the project deals with the authors' methodology and the findings about the medium's effects on students' thinking and writing processes.

The "Speaking to Write" researchers list reasons that student writers with physical and learning disabilities are often labeled "academic failures" (13), and view C-MOC as offering "an opportunity to reverse the cycles of failure that writing difficulties can perpetuate" (13). The project's materials are built around the idea that composing orally offers writers with disabilities a way to communicate effectively. The individual case study results support this outcome. The authors provide before-and-after writing samples to show how students produced more text (and more sophisticated, well developed ideas and sentences) after composing with VRT.

The "Speaking to Write" manual provides guidelines for training students to use VRT (including technical aspects). The authors urge readers to remember that speech recognition "is not fundamentally different from other modes of writing," and assert that the three "inter-related learning processes" that happen with C-MOC are "mastering the system, composing through oral dictation, and formulating written language" (32). Still, although oral composing is just another mode of writing, this mode changes a writer's way of putting individual sentences together. In an email correspondence concerning my study, "Speaking to Write" researcher Bob Follansbee commented on the complexities of oral composing:

   I do think the formal compositional processes are a bit more
   complicated now with continuous speech technology: the line between
   conversation and composition is that much less distinct. For at
   least some of us who do not produce well-organized text in normal
   conversation and who are inveterate tinkerers with our written
   language, SR poses some great challenges in producing writing.
   (For example, I backed and rewrote/typed the previous sentences
   several times before settling on my phrasing, and indeed, my
   point). ("Re: Dissertation on Oral Composition and Voice
   Recognition")

Follansbee's point concerning the blurring lines between conversing and writing leads back to an assertion made by Jane Danielewicz and Wallace Chafe. They assert that we must consider how to evaluate students' prose in which spoken elements make their way into the text. In an article seemingly prescient of C-MOC's complication of our notions of the writing process, they argue that students "carry over into their writing various practices which are less at home there, even though as spoken practices they may be quite normal and unremarkable" (213). The relationship between written punctuation and spoken rhythms is a complex one, and this relationship creates difficulties for novice writers (214). They conclude by suggesting that it may not be entirely negative for some spoken elements to transfer to writing. "Perhaps," they state, "punctuating as one speaks can in some cases lead to greater readability and greater impact" (225).

Follansbee's comments also bring to light the question of how to pedagogically bridge the gaps between "well-organized text" and "normal conversation" to help students compose through speech. If we embrace the advantages that writing by voice may present to our students, we must also embrace the ways in which this process may change their writing at the style level. The "Speaking to Write" authors highlight aspects of the mechanics of C-MOC that set it apart from other ways of writing. One concerns the need to constantly turn the microphone on and off (the commands for this are: "Go to sleep," and "Wake up"). They write that "for some students it may be distracting and interfere with his or her ability to focus on writing" (29). Thus, one difference between this and other forms of writing also stems from the added demand that students learn a new set of verbal commands for use during the writing process.

The authors suggest that when writers begin to use the software, they should begin with simpler writing tasks. Then, once the writer has developed some confidence with composing orally, the demands of the assignment may be increased (31). This training paradigm can perhaps also be applied to college composition courses or individual assignments in other writing intensive courses. The researchers provide recommendations for instruction that parallels most writing instructors' concern with content over mechanics. They warn that instructors should not overemphasize the problems associated with grammar and mechanics. Instead, instructors must support the increase in the quantity of writing produced as well as the work on higher organizational issues, "while helping students come to a gradual appreciation of the importance of writing mechanics" (99). Still, because college composition does tend to demand a quicker grasp of mechanics and syntax that increases with the level of the course, instructors who integrate writing into their courses may be more inclined to adopt a sort of hybrid process, in which students compose orally but edit by hand to eradicate computer-generated spelling or punctuation errors.

Conclusions and New Directions

Reviewing the findings from the three projects yields information about how C-MOC has worked with younger students, and it also provides ideas for how college writing teachers may integrate the technology into their courses. Instructors may start by having students dictate low-stakes writing tasks that do not encompass a large portion of the students' grades. These first low-stakes tasks may first ask students to dictate a letter or email to a friend on some rhetorical topic related to an ongoing project. Integrating C-MOC in this way would give writing teachers of writing greater freedom to introduce C-MOC at whatever point they wish to in the term. As the semester goes on, students may be asked to dictate longer project incorporating research and secondary sources. Because we realize that the writing process in college courses does often encompass research, it would also be important to have students use VRT to search the internet and library databases for sources relevant to their own projects. If the same benefits from these projects hold true for college writers, instructors may see an increase in students' attention to punctuation, voice, and format, as well as their ability to prewrite and plan their texts.

All three projects have shown that C-MOC alters the writing process and presents many benefits (such as speed, heightened awareness of formatting and organization concerns) for student writers. They also propose methods for exploring oral composing that can be adapted for student writers at the college level. But where do we go next with such inquiries? One suggestion is to study C-MOC in the context of college level writing tasks. And as we begin to embrace multi-modal composing across the curriculum (though such genres as blogs and webpages that incorporate elements of sound), students can use C-MOC to their advantage during the writing process.

Works Cited

Berieter, Carl and Marlene Scardamalia. The Psychology of Written Composition. New Jersey: Lawrence Erlbaum, 1987.

Bourdin, Beatrice and Michel Fayol. "Even in Adults, Written Production is Still More Costly than Oral Production." International Journal of Psychology 37.4 (Aug. 2002): 219-228.

British Educational Communications and Technology Agency. "Dfee/BECTA SEN Speech Recognition Project." 1 May 2005. <http://www.becta.org/uk/ page_documents/teaching/finalreport.pdf> 1-18.

--. "Speech Recognition Project: Project Information." 6 May 2005.http:// www.becta.org.uk/teachers/teachers.cfm?section=l_7_4&id=2707

Chamberlain, Michael A. Computerized Voice Recognition Systems and Their Application to the Mobility Impaired. Dissertation. Florida State University, 1993.

Communication Aids for Language and Learning Centre. "Introducing Speech Recognition in Schools Project." University of Edinburgh, 2002.

--. "Resources: Training Materials for Dragon Naturally Speaking." 6 May 2005. <http://www.callcentre.education.ed.ac. uk/Research/Speech_Recog_PRA/DragonSR_PRB/dragonsr_rb.html>

--. "Speech Recognition in Schools." 7 May 2005. <http:// www.callcentre.education.ed.ac.uk/Research/Speech_Recog_PRA/ speech_recog_pra.html>

Chafe, Wallace and Jane Danielewicz. "How 'Normal Speaking Leads to 'Erroneous' Punctuating." The Acquisition of Written Language: Response and Revision. Ed. Sarah Warshauer Freedman. Norwood: Ablex Publishing, 1985. 213-225.

Day, Sheryl Lee. Computerized Voice Recognition System Effects on Writing Skills of Community College Students with Learning Disabilities. Dissertation. Florida State University, 1995.

DeLaPaz, Susan and Steve Graham. "Effects of Dictation and Advanced Planning Instruction on the Composing Process of Students with Writing and Learning Problems." Journal of Educational Psychology 89 (June 97): 203-22.

Follansbee, Bob. "Re: Dissertation on Voice Recognition and Oral Composition." Email. 1 Sep 2004.

Murray, Donald. "Teaching Writing as Process Not Product." Cross-Talk in Comp Theory. Ed. Victor Villanueva. Urbana: NCTE Press, 1997.3-6.

Reece, John E. and Geoff Cumming. "Evaluating Speech-Based Composition Methods: Planning, Dictation, and the Listening Word Processor." The Science of Writing: Theories, Methods, Individual Differences, and Applications. Ed. C. Michael Levy and Sarah Ransdell. Mahwah: Erlbaum, 1996. 361-380.

Sara P. Pace, Lamar University, TX

Sara P. Pace, PhD, is an Assistant Professor of Rhetoric and Composition