Doing interviews with my new recorder, and taking pictures of sites I’m visiting with my trusty Fuji XE-1.
Interviewing Stuart Ralston, owner and head chef at Aizle, Edinburgh

The last several days I’ve been working through the hours of interviews I conducted on the road, editing reviewing and editing transcripts, and sending them out to people for review and editing and approval.

This has been work I’ve done my entire scholarly and professional career. When I was researching my senior thesis, on electrical engineering at MIT between the world wars, I spent many nights sitting in my summer rental on the edge of the MIT campus, listening to interviews on a cassette tape, transcribing them on a typewriter (a couple of the interviews are in the MIT archives). This is what I did instead of hanging out.

It’s slow work, but it’s essential. Reviewing the transcripts let me get reacquainted with conversations, identify those parts that I’ll later want to quote, and look at the overall structure of the conversation to see if there are themes that people bring up that I hadn’t noticed before.

Of course, they’re all answering questions that I ask, but it’s also important to listen for the things that people bring up on their own. Given how much of my work relies on interviews, this is a step I can never afford to ignore.

Fortunately, I recently stumbled on a new transcription service called Otter.ai (I believe they’re located somewhere here in Silicon Valley).

Scenes from Shoreditch

Otter is pretty good at identifying specific words (though it does make systematic errors, particularly when it’s dealing with Scottish or English accents), but it struggles with sentences, and paragraphs are hopeless. But I don’t think these are flaws in the program; rather, it reveals is how people hardly speak in sentences, and almost never speak in complete paragraphs. I have to decide where one thought ends and another begins, and make a decision about whether to start a new sentence, or create a new paragraph. That’s all editorial judgment on my part; rarely is it something signaled by the transcript itself.

The good news is, the interviews are as detailed as I remember, I managed to get good recordings of all but one (damn 2 GB card), and so it’s all worth that’s worthwhile, and will go a long way to letting me finish what feels more and more like my most important book yet.