'Computer, Listen to Me About My First Draft'

Remember the first time you tried to type? I do. My fingers felt the size of German sausages. Unwanted letters appeared on the blank sheet of paper in my typewriter.

No matter how hard my typing teacher urged me to look at the page, I could not keep myself from looking down to watch my fingers on the keyboard - not that it helped my accuracy.

I've never completely mastered typing. Not even the advent of high-tech word-processing programs (which at least made it easier to correct my mistakes) has made the task more palatable. While I have progressed beyond the two-finger stage, I'll never break the 30 to 40 words per minute barrier.

Which is why I like the new voice recognition software. Basically, I talk to my computer and my words appear on the screen in front of me. (Well, not all my words, but more on that later.) The only reason you'll need your fingers is to twiddle your thumbs while you're thinking about your next pearl of wisdom.

Voice recognition software has come a long way in the past year. Earlier versions were expensive (often between $700 and $800 as recently as two years ago) and required the user to slightly pause between each word, which produced a clumsy, uncomfortable staccato rhythm of speech.

But two companies have recently developed programs that eliminate the need to pause, thus allowing "natural" speech. The programs are IBM's ViaVoice ($99 - you've probably seen its TV commercial about the man who won't let his fellow employee "talk" to his machine) and Dragon System's Naturally Speaking ($129-$149). It's this later version that I've been using. (Microsoft is also developing a similar program.)

Actually, "using" is the wrong word. "Training" is more to the point. The more you "train" the software to recognize your personal inflections and speech patterns, the more useful it becomes to you. That's why patience is important, if you want to use this software. More than once I've reflected on the similarities between training my computer to recognize how I say "online" (which appeared as "a line" or "only lines") and helping my two-year-old son learn to say "snake" instead of "shhhnake."

There was another aspect of learning to use the program for which I was unprepared - the difference between "writing" and "talking." Think about the way you write a letter to your mother. Now think about the way you talk to her on the phone. Talking may be easier, but it is almost always more convoluted. Writing or typing requires a formality that talking doesn't.

Dictating punctuation, for instance, was a real challenge. Nor do you type "uhms" and "ahs." So when I first "talked" a column, it felt unnatural. Voice recognition software also won't help much, if you work in a noisy environment. It translates stray sounds into words. Sometimes this can be quite amusing. The first time my program picked up a loud door slam over the mike that comes with the program, it translated the sound it heard as "pretzels."

Voice recognition software is not likely to completely replace writing (or typing for that matter), no matter how accurate the software becomes. Still, it is an enormously useful tool for writing first drafts.

* Tom Regan is supervising online editor of the Monitor's Electronic Edition http://www.csmonitor.com