I cannot beleive I still haven’t written about seeing the aurora in 1988. Maciej’s long piece on his stopover in Iceland calls it to mind. I really want to write about it seriously but I have a specific method in mind to pursue. I want to use voicerec for the first draft without correcting until I’m done with the draft.

Unfortunately, I have been forced to implement my main desktop box as the server for the moment, which means I don’t have convenient access to my installed and trained copy of ViaVoice.

I did use ViaVoice to begin some of the transcriptions – listen to tape, say what you hear, correct, repeat – but ViaVoice, while ostensibly designed to encourage non-dictation style speech, really does better with slow, deliberate, clearly pronounced words rather than rapid, conversation-speed speaking.

So I’ve been transcribing by hand, which is not so good for the wrists and forearms.

So far, normal conversation clocks in at about 10,000 an hour. Ten thousand words! The Fagles translation of The Illiad looks to have about 199,000 words – twenty hours of speech. A highly productive writer will produce about 10,000 words a day – which means roughly eight times the amount of work goes into writing over speaking, I think.

To date, I’m able to transcribe about 5,000 words per day – but of course, my day, here at home is both more focused and shorter than any work environment I’ve ever had. It’s shorter because I’m both lazy and a night person, which makes it difficult for me to rise in the morning, and when I do, it’s rare I focus on anything until noonish. It’s more focused, because of course the twenty-first century American workplace is fraught with productivity boosters such as meetings, telephone calls, voicemail, and coworkers. I’m guessing it’s a wash, and that would be my output in a work environment was well.

One of the striking things about speech, observed in transcription is the way we insert mouth music into our sentences, drop sentences and clauses uncompleted, and create meaning in casual speech by piling up approximations until we see that our conversation partner has constructed their own understanding from our troweling of parts and participles, hurried, organic, held together with ‘uh’ and ‘um’ and ‘you know’ and ‘like’ and ‘and’ and ‘I mean.’

Believing myself not to be a fiction writer, I have never even wondered if I could write dialog. Watching the words collide and slip into one another and stick or slide or stop is an education, currently fascinating me.

What concept, actually, ‘writing dialog.’ I believe I’ll paint some looking, or perhaps sculpt some motion.

This interest is not the impetus for using the voice recognition as an experimental composition tool, though – it’s more like writing your essay on a Newton and deciding beforehand to not correct the interpretations. Of course ViaVoice is waaaay better at it than the Newton. So it’s more a way to force different word choices into the prose.

I’ve also experimented with both Word’s built-in summarization tools, which will select key sentences with the intention of preparing an executive summary, as in the initial sections of a business plan or even in certain kinds of technical writing. As such, it’s fairly graceless but can be useful for cutting to a certain length. As far as cutting from 600 words to 250, mmm, not so much.

There are some online summarization tools as well; so far I haven’t found any that work quite as well as Word’s. Heh. So, uh, thanks, Redmond! can I get you to add a real, fullblown grep to search if you ever see fit to make a new Word for OS X? If not, well, no harm done.

One thought on “Figuring speech

  1. Well sure, yer gonna have trouble if you use the Fagles translation of The Iliad. I mean, Jeez, the Fagles? Why not just use the Sprague? Trust me…use the Fleagles translation. We’re talking six, seven hundred words, tops.

Comments are now closed.