Transcribing is not impossible to learn. I had no experience prior to this, but I think I'm doing fine. But transcribing is
intensive. I personally have a limit of about 5 minutes transcribed content/30 minutes of transcribing before I have to take a break. Your mileage may vary of course, but if you want to try it out, don't fret. Transcribing is also fun.
When transcribing, you have two somewhat conflicting goals: (1) Make the text as similar to what is said as is possible. (2) Make that text as readable as possible.
The first goal is basically about not wanting important details to be left out of the transcript. The DF Talks are also (sort of) official statements, and often quoted to spread information on DF's development within the community. These quotes need to be accurate, and listeners and readers should agree on the content of the talks. My personal philosophy is that the "style" (e.g. figures of speech) should be preserved as well; ideally, the transcriber should leave no trace at all, but being in the credits of course.
The second goal is equally important, especially since a fair portion of the DF talk "consumers"
only read the talks, without the audio to back them up. The experience should be pleasant. The dilemma is, how much should you edit the spoken word to make it readable, as intonations, pauses etc. can be difficult to transfer directly without becoming a mess of half sentences and stuttering.
Here are some notes I derived from DF Talk #19, the last one mallocks transcribed (and thus presumably their best). First come some methods I am mostly positive or neutral on, and then some I personally don't really like:
General notes on mallocks' methods:
- '' instead of ""
- "Yes" instead of "Yeah" at the start of Toady's lines (but this is not consequent)
- Contractions are allowed, both really common ones like "that's", "it's", "don't", but also "you'll", "we've" etc.
- "gonna" => "going to"
- " ... "
- Sentence-start clusters are omitted; even short ones: e.g. "But yeah so" => "So".
- "And", "But", "So" may simply be removed from the start of sentences.
- When a sentence is revised mid-sentence, the parts that don't fit the revised version (gramatically) are omitted.
- Most "sort of", "kind of" are omitted. They are only used where there is uncertainity, not those used as padding, while thinking etc.
- Some "just" are omitted.
- "or something", "and so forth" are generally NOT omitted
- ; allowed, may replace "because"
- Half sentences cut off by the other person are omitted, unless that's what the other person is responding to.
- Repeated whole sentences are shortened to only one: e.g. "The hole is still there. The hole is still there." => "The hole is still there."
- Laughing is not marked.
Methods I personally dislike:
- Some sentences at the end of a "block" are omitted.
- Short responses are generally omitted. (note that Rainseeker is more "chatty" than Capntastic, it miiight be justified)
- Simply too harsh editing, some revised sentences contain information not in the revised version. Mallocks methods removes this information.