by Charles Lowe 
Florida State University  

Currents in Electronic Literacy Spring 2001 (4), <http://www.cwrl.utexas.edu/currents/spr01/lowe.html>  

Introduction 

1. Last spring during a writing about popular culture class on Star Trek, my students taught me what at first seemed an insignificant piece of trivia. As many fans of the series know, Majel Barrett Roddenberry is an important part of the Star Trek universe. Not only was she wife and advisor of series creator Gene Roddenberry, but she also starred as a member of the cast in Star Trek: The Original Series (1966), later continuing this tradition with a new character in frequent guest appearances for Star Trek: The Next Generation. When our class read an article of hers describing Star Trek's humanistic vision, my students quickly pointed out a little, less known fact about her role in Star Trek: Majel had been the voice of the ship's computer throughout all four Star Trek series. 
2. While this piece of trivia may seem meaningless to most, for me it recalls my earliest awareness of speech recognition technology. As an adolescent during the 70's watching the reruns of Star Trek: The Original Series , I always marveled when the computer "listened" to the instructions of the crew and responded back with Majel's voice. Meanwhile, contemporaneously with Star Trek, the Stanley Kubrick and Arthur C. Clarke masterpiece, 2001: A Space Odyssey (1968), featured a dramatic juxtaposition of music, silence, and spoken communication between man and machine based intelligence.  
3. With 2001 finally upon us, Clarke's artificial intelligence HAL, along with a manned mission to Mars and cold fusion, seems a little less the dream of science fiction, although not yet science fact. Research in computer artificial intelligence has yet to construct a reasoning machine, but today's computer scientists would say that the computer "communicates" with the Star Trek crew using speech synthesis, natural language processing, and speech recognition. In fact, long before either computer scientists or science fiction writers envisioned computers that could hear and speak, Alexander Graham Bell expressed the desire for "a device that would make speech visible" for the hearing impaired (Klevans 3). As was his wish, at the research institute which bears his name, Bell Laboratories, in 1952 Davis, Biddulph, and Balashek constructed the first "voice input system" which recognized "ten digits spoken by a single user with 100% accuracy" (Fournier 2). Further advances from that original success in voice recognition have led to AT&T's well-known collect calling services. During a collect call, a mechanized voice asks the recipient of the call to answer "yes" or "no" as to whether or not to accept charges. It takes little to extrapolate from this one example of human/machine voice communication to conceive of a near future where speech recognition as a means of textual production might become more than probable.  
4. However, the example above is more correctly described as voice recognition and not true continuous speech recognition -- the computer only has to differentiate responses for a vocabulary of two words. With continuous speech recognition, the computer must be able to recognize vocabularies of thousands, or hundreds of thousands, of words, discerning when one word ends and another word begins in situations where pronunciation of utterances varies from user to user, and to identify words and speech even when users might pronounce words in various ways. Hence, it's no wonder that the early so-called speech recognition programs available in the 1980's seemed limited at best. The user needed extensive training with the program before initial use and had to cope with limited vocabularies far less than that necessary for normal spoken conversation. Because of the difficulty in recognizing word boundaries, the user was also required to insert a 1/10th of a second pause between each word in order to break the flow of ordinary conversation. 

A Look to the Past: An Aid to Writers with Disabilities

 
5. Considering the expense of the software due to large development costs passed onto the consumer, as well as the price of early PC's, the economics of using a speech recognition system made it a viable technology for only those with the greatest need, those who would not be able to produce printed text without it. Speaker independent voice recognition software with speaker dependent customization and larger vocabularies ran upwards of $5,000 (Chamberlain 45-46). (1) For example, even in 1993, Dragon Dictate 2.0 had a 25,000 plus preexisting vocabulary and 5,000 word user definable one for a price of $4,995, excluding the 386/SX or DX IBM compatible PC necessary to run the software (Chamberlain 46). Additionally, the software limitations of these earlier systems made them a bad alternative to keyboarding for most individuals, since dictation speeds per minute and accuracy ratings were well below those of the average typist. Dragon Systems claimed "average dictation speeds of 30 to 40 words a minute" with accuracy "in the low 90's in percentage terms" for the Dragon Dictate 2.0 (Chamberlain 83). 
6. As an expensive and inefficient mode of text production, speech recognition was deemed only practical and desirable, yet beneficial, by those forced to look for alternative means of textual production beyond keyboarding or handwriting. In the 1990's, research at Florida State University generated two dissertations which posit the benefits of speech recognition for those with disabilities or impairments. The first and most informative, Michael A. Chamberlain's Computerized Voice Recognition Systems and Their Application to the Mobility Impaired (1993), describes the evolution of speech recognition technology, explores the commercially available products at the time of the study, and conducts a case study of eighteen mobility impaired individuals who lack the physical capability to use the typical computer input interface, i.e. the mouse and keyboard.  
7. Of these three areas, the most significant conclusions arise from his case studies of individuals with repetitive stress injuries, multiple sclerosis, muscular dystrophy, fibromyalgia, and paralysis: "In 16 of 18 case studies, subjects said that computerized speech recognition had made a real difference to the quality of their lives, and the two others said, 'it hasn't yet, but it probably will'" (Chamberlain 201). Additionally, all members of the case study "reported that using computerized voice recognition systems had increased the amount of communication 'messages' they sent" (Chamberlain 202). Chamberlain rightly concludes from his data that mobility impaired individuals using speech recognition have become more active participants in society through their ability to communicate via email coupled with the opportunity to return to the work place or college with the ability to produce text. Psychologically, for the individuals in the study, speech recognition has led to richer, fuller lives by reducing feelings of inadequacy and isolation. 
8. The second dissertation from Florida State University, Sheryl Lee Day's Computerized Voice Recognition System Effects on Writing Skills of Community College Students with Learning Disabilities (1995), also provides evidence of the benefits of using speech recognition technology. Students with learning disabilities often have trouble getting language onto paper because of problems such as "spelling, grammar, punctuation, organization, and coherency" (Day 4). However, speech recognition would appear to circumvent many of these difficulties. Using holistic ratings and examination of text features within the writings of four community college students, Day found: 

that dictation as a method of text production resulted in two of the three students writing essays judged to be a better quality than essays written with a word processor when rated holistically by two independent raters  . . . data indicate computerized voice recognition can assist some students with learning disabilities to compensate for written language difficulties. (121) 

9. When considered together, Day and Chamberlain's studies illustrate that making speech recognition technology available in schools, universities, and the workplace could help individuals who feel incapable or disadvantaged in such environments become better able to participate in writing activities which most people take for granted. In an era in which we question the effects of computer technology on literacy, it is promising to find a positive benefit: speech recognition may offer increased access to members of society previously marginalized by limited written communication skills. 

The Future Becomes the Present 

 
10. When looking back over the incredible technological advances during the digital revolution of the last twenty-five years, most of us can see, with or without the help of science fiction visionaries such as Roddenberry and Clarke, that software and hardware design should soon make improved versions of speech recognition economically feasible and consequently more accessible to larger portions of society, instead of only as a forced choice option for those without other means of creating text. One can easily imagine the following scenario in the not too distant future:  

Following dinner, Megan prepares to write a short report due the next day for her 5th grade social studies assignment. After spending about an hour on the computer, she walks away with her report in hand, having discovered the information she needed fairly easily on the Internet. However, Megan differs from today's students in that her hands never touched the keyboard or mouse in preparing her text; the computer was controlled and her text was created with the sound of her voice. 

11. What many people are only just starting to realize is that today, with a moderately priced computer (under $1000) and less than $100 in software, the creation of text through voice is now a reality; Megan's brief narration above no longer belongs to the realm of science fiction, nor is it wishful thinking. Word processing with speech recognition capabilities is within financial reach of the average American family, for in 1997 Dragon Systems released the first continuous speech recognition program capable of dictation speeds of upwards of 160 words per minute. (2) 
12. Admittedly, most of us who are already comfortable with the keyboard may never switch over to speech recognition as a method of generating text. I personally have found the act of composing via speech disconcerting, although this seems a reasonable response given the hundreds of hours composing by handwriting and typing which has conditioned me to write without speech. Generations in our culture, like myself, already educated under the influence of pre-digital print literacy may never switch over to speech recognition as a means of textual production, unless forced to by development of chronic disabilities such as carpal tunnel syndrome or arthritis. Then speech recognition may bring a welcome relief from the pain of typing and using the mouse.  
13. But we must remember that digital literacy offers new standards, and it may only be a matter of time before speech generated text becomes an indication of high digital literacy. New generations of students already proficient in uses of technology (which many of us had to acquire as adults, either willingly or reluctantly, but generally with more difficulty than if assimilation of the new technology had occurred earlier in our lives) will not be as predisposed to keyboarding. As Walter Ong explains, "oral speech is fully natural to human beings" whereas "the process of putting spoken language into writing is governed by consciously contrived, articulable rules" (82). For them, moving the fingers across the keys may seem extremely unnatural and time consuming compared to speaking to the computer. Children learn to speak not long after they can walk; typing is an artificial form of communication created by man, and one which they may decide that they can live without. All of us who consider ourselves accomplished typists, these days otherwise known as keyboarders, remember the painful and frustrating number of hours required before satisfactory levels of ability were reached. Ruthann Dirks and Marvin J. Dirks's study "Introducing Business Communication Students to Automated Speech Recognition: Comparing Performance Using Voice Input and Keyboarding" (1997) questions whether educators should continue teaching keyboarding; their study indicates that speech recognition may be easier and faster to learn. After completing a 1 ½ hour tutorial using DragonDictate, 20% of the students achieved higher speeds with the dictation software than through typing (Dirks 154). Significantly, DragonDictate is an older, discrete speech based version with much less accuracy and speed than current systems; 20% may be an extremely conservative figure for the results which might exist were the study conducted with current software and hardware.  
14. It's not hard to imagine that school children might choose composing via voice over keyboarding. Writing is an alien, mechanical means of recording language, a medium requiring a conscious physical exertion during composing in the early stages of acquisition. Speaking to the screen should seem more natural to almost anyone without the investment of training and practice in handwriting and keyboarding. While I must admit that composing through speech seems somewhat awkward since I have internalized the rules and effort required to make writing automatic, I still continue to work with the software on occasion, intrigued by the idea that I might one day have the ability to speak at over 100 words per minute and have text appear on the screen. For future generations, though, this decision won't be a matter of choosing whether to acquire speech recognition skills after learning to type: as former keyboarding teacher Patrick J. Highland suggests, keyboarding may become a more specialized skill, "used more for editing and less for initial input of a document" (Highland 32).  
15. Whether compositionists choose to explore this new method of textual production or not, computer manufacturers will continue to develop it for the marketplace. Software developers foresee a future in which more user-friendly computers will be controlled only by speech. As well as producing text, newer speech recognition programs already allow the user to control most Windows applications, including Internet browsers and email. Meanwhile, word processors and other common applications no longer benefit from increases in processing power. Because hardware manufactures depend upon advances in software design to fuel the need for better hardware, cpu-intensive technologies such as speech recognition and natural language processing continue to strengthen the symbiotic relationship between software development and hardware advance. I predict that it won't be long before speech recognition software comes free of charge on every Dell, Gateway, or Compaq sold in an effort to inculcate a new consumer base who will be partially or completely dependent on speech recognition for textual generation. Fully recognizing the demographics of their market and realizing that potential users need to be trained young, Dragon Systems, the leading manufacturer of speech recognition technology over the past 20 years, includes speech models specifically tailored for children nine years old and up as well as adults within their latest versions. While the parents may never use it, their children will delight in the magical product which allows them to create text on the screen without keyboarding. These same children will bring assignments home from school and find writing through typing tedious, while speaking their essays onto the screen an ease and a joy. 

Implications for Composition: Revisioning Speech and Writing 

16. As composition specialists highly trained in the art and skills of writing, most of us are content in our ability to generate text through keyboarding and may even feel threatened by the introduction of oral discourse into the composing process. Nevertheless, in a recent article in Kairos, Stanley Harrison points out that we must become involved in the identity construction of the "SoundWriter cyborg," a "machine/organism system" featuring "a speaking human," "a microphone," "a voice-to-text computer program," and a "text-to-voice program" (2). While acknowledging the "incredible and, heretofore, unexpressed potential" of the SoundWriter, Harrison warns that "corporate, popular, and educational communities have brought a consistent and narrow set of concerns to bear upon our understanding of the ASR/human interface" (2-3). Yet, compositionists, "who have, heretofore, been silent on the matter of SoundWriting," must become involved in this identity discussion.  
17. While I am less concerned by the fearful tone that Harrison's essay raises as to the hegonomic domination of this new cyborg construction, I do believe that compositionists need to begin considering the influences that this method of textual production (3) will have upon our students and the way in which it requires us to revision composition. Whether we see the digital and electronic age as a move toward Ong's secondary orality or not, we need only recall the impact of word processing on the writing process to glimpse the magnitude of the impact which speech recognition as a means of textual production may have upon both the field of composition and definitions of literacy, now and in the future. To fully understand the implications and potentials of speech recognition, we need to move beyond the threat that this technology poses for the status quo of print literacy and the historical primacy of writing over speech that Jacques Derrida has sketched out in Of Grammatology and grant textual generation through speech an equal consideration in composition theory.  As we create syllabi for writing classes, read composition journals discussing old and new theories, and devote hundreds of hours each year to commenting on and evaluating student papers, we should acknowledge our natural biases resulting from our constant involvement with and love for writing, and instead see each media as different, but equivalent. 
18. Consequently, I would suggest attempting to mediate these boundaries imposed by our predilections for one medium over another by looking to the work of linguistic studies of the 1980's as a lens for reexamining conceptions of speech and writing. For example, in response to naive comments about speech as inferior to writing, M. A. K. Halliday draws upon his experience with spoken discourse in pointing out that some speech uses extremely complex sentence structures (58). In expanding on his initial observations, Halliday's study demonstrates that written text has greater lexical density--the "proportion of content words to total discourse"--and thus we might incorrectly assume that "writing is more complex, since presumably lexical density is a form of complexity" (62). Yet, under certain circumstances, such as spontaneous spoken discourse, speech demonstrates greater "grammatical intricacy" than writing (65-66). This complexity depends upon the type of discourse, just as with writing, and therefore we should "question the assumption that written language is syntactically more complex" (66). 
19. In continuing to establish the equivalency of speech and writing, other linguistic studies explore the differences between the two. Deborah Tannen points out a "recurrent hypothesis" explaining the relationship, or rather differences, between writing and speech (3). Whereas "spoken discourse is highly context bound," on the other hand, "writing is decontextualized" (3). Tannen explains that while this hypothesis is informative, conclusions here are generated from the differences in "the genres selected for analysis--casual conversation on one hand and expository prose on the other" (3). Similarly, Halliday's conclusions agree by pointing out that previous assumptions of speech as more flawed, more simplistic, may be based upon inappropriate models where speech is consciously planned: the speaker, listening as he speaks "to check the outcome . . . naturally" makes the speaker "lose [his] way: to hesitate, back up, cross out, and stumble over words" (68). On the other hand, in spontaneous discourse such as conversation, "the clause complexes tend to flow smoothly without you falling down or changing direction in the middle" (Halliday 68). Martin Nystrand writes that it is a fallacy to distinguish "written and spoken language in terms of the autonomy of the text. . . . cohesion results not when language is written but rather when language is put to particular uses, especially those uses which bridge discrepancies in writer-reader knowledge, as in expert-layman communication. Language is not composed because it is internally cohesive" (201). He concludes that "it is clearly a mistake to associate the spontaneity of casual talk with fragmented expression, and equally wrong to confuse elaborateness of text with fullness of meaning" (211). 
20. Furthermore, linguists have redefined the differences between writing and speech through a clearer understanding of cohesion. Horowitz and Samuels explain that "cohesion is expressed through dexis (referring to items outside of a discourse or text) but also through prosodic cues (the pitch, stress, and pauses expressed in language) and paralinguistic devises (such as facial expressions, lifts of the eyebrow, smiles or frowns, or body language such as pointing or distancing oneself from a listener)" (7). Tannen similarly agrees with this description of spoken discourse while defining writing as having "lexicalization and complex syntactic structures" which "make connectives explicit" with "subordination and other foregrounding or backgrounding devices" establishing "relationships between propositions" (3). One would then conclude that "oral strategies . . . make maximal use of context by which maximal meaning and connective tissue are implied rather than stated, and "literate strategies" are "those by which maximal background information and connective tissue are made explicit" (3). 

Applying Speaking to Invention: Freespeaking

21. Once we can see speech as an equivalent, although different, discourse -- and in doing so, recognize the close parallels between speech and writing -- it becomes possible to imagine more uses for speech recognition in composition; for example, its potential application to invention. In one of the few composition studies that links oral discourse to effective composition, Christian Koch argues that common invention strategies such as "start now and keep going; write in chunks; start anywhere; don't worry about language and mechanics while you write; write more about your focus" are no more than "written versions of behavior typical of speech" (64). Koch sees these oral strategies as "content generators," while also noting that spoken discourse is "powerful stimulus" to the production of text, and because "we cannot revise what we have said," we must "go on producing words until we feel that we have expressed the idea, or solved the problem, that we had in mind" (68).  
22. Indeed, Koch's work is more than just important for breaking down the boundaries between speech and writing, or for understanding how oral strategies are implied in traditional and current invention strategies. These thoughts also can help us reread current pedagogical theories about invention, readings that suggest positive implications for using speech recognition for content generation during early drafting. In summarizing Ken Macrorie's feelings, Erika Lindemann explains that freewriting "produces honest writing, writing that is free from phoniness or pretension. The writer must write fast enough to use 'his own natural language without thinking of his expression'" (Lindemann 58). Assuming that natural language occurs in thought at the same speed of spoken conversation, neither handwriting nor typing are fluent enough to produce this honest flow of writing. Additionally, because the writer is engaged in a text production process when typing or handwriting, cognitive processes may reimage thought into an appropriate form for writing, even when following Peter Elbow's advice to " simply write" and to never stop to think about it (qtd. in Lindemann 58). Writing without thinking is contrary to every other form of writing that we do. During composing, we think about we are going to say, and watch to see if the sentences are formed correctly, spelled correctly, logically sensible, and well-organized. When freewriting, we must consciously control these urges to compose text. In spoken conversation, most of us do not compose as we speak with the same cognitive processes--I, for one, have often been guilty of speaking without thinking. Speech recognition frees the user to speak as in conversation with the spirit of Elbow's theories of freewriting never possible before.  
23. Let me suggest then that freespeaking -- applying the concepts of freewriting to the generation of text with speech recognition -- may offer increased potential over freewriting in content generation during writing. Consider how freespeaking would seem to better satisfy some of these tenets of freewriting expressed above and in Pat Belanoff, Peter Elbow and Sheryl Fontaine's definition in their introduction to Nothing Begins with N. As they explain, in spontaneous spoken discourse we do not often "think about spelling, grammar, and mechanics" (xii); additionally, because this is a natural way of talking, a writer new to freespeaking might acquire this method more easily, whereas the freewriter may be rebelling against her/his sense of writing as a polished, constructed piece of text. And, freespeaking may decrease the tendency of freewriting to not "make sense or be understandable" (xii) since this quality of freewriting may occur as the writer attempts to transfer thought into text using the mechanical means of handwriting or typing; some of the confusion in the freewritten product may be more of a bottleneck problem, a lack of bandwidth for this transfer.  
24. Recent studies of freewriting would also suggest other potential benefits for freespeaking over freewriting. Fontaine states that even when encouraged to be exploratory and ignore the rules of written discourse, that students remain "close to common convention" (10). She further explains that this may be in part because we have succeeded so well at teaching students discursive structures that "they cannot write without using them to organize their thoughts" (11). I can imagine that using freespeaking may give students that extra freedom to generate content, a liberation from the restrictions imposed by years of structural approaches to composition based upon print literate strategies.  
25. Freedom through freespeaking may also arise through the rhythms present in the free flow of thought which are often carried over into spontaneous spoken discourse. Ken Macrorie writes that he often hears and reads of "an emphasis on the need for student writers to pick up the rhythms of ordinary speech in their writing. . . . that powerful rhythms can't be planned for sentences but usually arise from meanings that count for the writer (186). Are these rhythms transcribed onto the page as if by magic, or is there another way? Macrorie assumes that "rhythms, metaphors, analogies, powerful sound effects, brilliant connections between ideas and objects: all these at their best are given by the unconscious more often than they are planned and forecast and contrived" (186). Given the supposition that such prose cannot be taught as if by skill, even student writers, rarely the masters of written discourse, might produce these qualities more often when composing initial drafts through speech, their normal and more familiar medium of communication.  
26. Similarly, in "The Shifting Relationship between Speech and Writing" when discussing nine reasons "why writing needs to be like speech," Elbow admits that "speech" has "a magic that writing lacks--call it presence, voice, or pneuma--but the truth is that we tend to experience meaning somehow more in spoken words than written ones" (298). While some of the magic of speech may be tied to paralinguistic communication and through a speaker's attention to the perceived audience, it's quite possible that some "magical" meaning making may occur during the intensity of spoken discourse. And, like Macrorie, Elbow feels "the best writing has voice: the life and rhythms of speech. Unless we actively train our students to speak onto paper, they will write the kind of dead, limp, nominalized prose we hate--or say we hate" (291). Notice here that the italicized words are Elbow's notations, not my own, emphasizing the importance of spoken discourse in composition. Also, Elbow points out the common assumption that "the function of writing is to record what we have already decided--not to figure out whether we believe. . . . If we were speaking, we would be more likely to speak the train of thought as it comes to mind even though we're not sure of our final opinions" (287) Later, he compares "spontaneous writing" to "freewriting," noting that they are both writing "in which we put down whatever words come to mind--without regard to the conventions of forming words and without regard to the quality of the writing" (290).  
27. Other considerations may enlighten pedagogical considerations as well, suggesting that applications of speech recognition as invention in the classroom may benefit developmental writers. While studying the prewriting techniques of 11th grade students, Carol Pope and Doris L. Prater found that students preferred freewriting over "outlining, looping, clustering, or cubing", although "advanced and average groups gave freewriting a high ranking, [while] basic students gave only an average ranking to this strategy" (67). I would conclude that this is most likely because the mechanics of writing inhibited the process--thus the reason that the basic students "ranked talking as their highest rated strategy" (Pope and Prater 67). Also, Pope and Prater point out a need to "convince basic students of the value and worth of" prewriting strategies, since they used these heuristic tools less than the average and advanced groups (70). Isn't it possible that since the other tools involve writing, which basic writers struggle with, that more basic writers find less advantage in using them, that they doubt the benefits of invention as we define it? Freewriting requires tearing down the inhibitions, letting oneself go free to generate the content within the mind. How can unskilled writers to whom the tool is not second nature reach a level of comfort where they are not struggling with the mechanics of writing?  
28. Could the tendency of basic writers to create simple sentence structures result from difficulties with the mechanics of creating text, and not an inability to form more complex syntactical language constructions in their heads? Halliday's observations of complex sentences in spoken discourse might lead us to conclude this. The process of transferring thought to writing, the necessity of slowing down, might cause students to have to condense thoughts into generalizations, or even lose their ideas altogether. This is why we advise them to freewrite and worry only about generating content in early drafting. For some students, the mechanics of typing or handwriting may be the only stumbling block; on the other hand, as Reece and Cummings point out, when using computer "speech-based composition" the writer can compose at the rate of speech instead of handwriting or typing (264). Thus, students composing orally may end up with greater detail, greater sentence complexity, and more text, a theory supported by Reece and Cummings's work. Their study concludes that speech-based composition of young students using a listening word processing (4) (which allowed them to see the results on the screen as they typed) produces better writing than with standard dictation or normal writing (380). 

Conclusion 

 
29. Even given doubts that freespeaking may be more beneficial to textual production than freewriting, the possibility exists that speech recognition may have increased benefits for generating a stronger voice, better and more content, and a superior writing style. As Halliday concludes, "speech and writing will appear, then, as different ways of meaning: speech as spun out, flowing, choreographic, oriented toward events (doing, happening, sensing, saying, being), processlike, intricate, with meanings related serially; writing as dense, structured, crystalline, oriented towards things (entities, objectified processes), product like, tight, with meanings as related components" (80). At the very least, speech recognition may invest the text with desirable elements of orality which writing sometimes has difficulty emulating, a minimal justification for compositionists to begin thinking about this new method of producing text.   
30. Regardless, the choice of whether or not to use the new technology is not ours to make. Considering how quickly students have assimilated the tools of electronic discourse during the past decade, we don't have to look to science fiction writers such as Roddenberry and Clarke to imagine a ubiquitous use of speech recognition by the middle of this century. Without consciously or unconsciously precluding the superiority of the written text, compositionists must see spoken discourse as a rich source for generating content during invention work while concurrently expanding our comprehension of the composing process. If nothing less, this essay should indicate a need for research in understanding spoken discourse composition, both to prepare us to better understand freespeaking and the newer generations using speech recognition. Such students will have closer ties with orality in their sense of what writing is and where it comes from. Their shift in perspective will become our new paradigm for composition: writing as a process will become discourse as a process, and compositionists will become comfortable teaching and discussing the development and evolution of oral, written, and hybrid texts.  

And, assuming that speech recognition will enable students to generate better content during invention via freespeaking, if we begin now, we can concentrate our efforts on other compositional problems, for, as Rob Enderle, director of desktop and mobile technology at Giga Information Group positively forecasts, "In five years[,] you'll look back and wonder why you ever used a keyboard to type" (Randall 2).  

Works Cited