Starting at the point of having video recorded on a camera, my workflow tends to look something like this (though it depends a bit on what my research goal is):
- Use MPEG Streamclip to capture and compress the video from the camera
- If the video is of an interview, transcribe it using Transcriva (a similar program for Windows users is Inqscribe)
- Export the transcript as an RTF file
- Import the video (and transcript if its an interview) into Transana.
- If the video is of observational in nature, create a content log (as a kind of fake transcript) with time coding in Transana. This or a detailed transcript allows clips to be identified and categorized
- Usually at this point, I create keywords for structural characteristics such as which participant is speaking and the materials involved. I then go through the video and identify where these characteristics exist. This gives me a way to break down the video and if I have a lot of videos, it allows me to search across them for similar instances.
- Using the clips and keywords functions in Transana, I identify and organize instances across the video data I have. I tend to work in an inductive recursive way and as I create new categories and keywords, I go back through the instances I have already identified.
- Once I have identified specific short sections of a video or videos that I want to perform more detailed analysis on, I the move to ELAN
- In ELAN, I usually create multiple channels and input as much timecoded detail of the interactions in a clip as possible. The might include:
- A channel for the discourse of each speaker
- A channel for gestures
- A channel for physical interactions with the technolog involved
- A channel for the communicative actions made by the technology