At the end of the last exercise we wondered whether we should be putting down descriptions like "holds gaze while scratching cheek', or "door handle opening". If they are to be useful, they have to be reasonably accurate, impartial and complete.

But unfortunately we can't be utterly confident. Or, at least, not as confident as we can be about having captured the spoken language.

Our wordy descriptions of body movements (including the movements of eyes, hands, face, arms...) is crude. It is not in the same league as the precision of spoken language, where small nuances of difference have a corresponding notation. We can write a given sound-stream down as "The bats are flying", or "The rats are frying", or "The cats are crying" according to the fine distinctions our ears make. But, aside from deliberate systems like British Sign Language, our spoken language doesn't find it easy to equivalently encode fine distinctions of bodily movement. (Think of writing [holds gaze while scratching cheek] - is that as precise as reporting Lyn as having said "they're in the kitchen"?).

A more subtle danger is that when we describe bodily movement, we often use a word that makes a guess at what the person is thinking or feeling, or what they want to achieve. For example, suppose we wrote down that Zoe 'flounced' out of the room, or 'walked out smartly', or 'escaped' from it. The problem is that each is a guess at what Zoe is 'communicating'. It's short-circuiting the whole point of the exercise, which is to describe something neutrally, so that we can then analyse it dispassionately.

A third danger is that we might think we have a complete account of all the relevant body movements, and proceed with the analysis as if that was all there was that was relevant. But even though the transcription has now blossomed to over 50 lines, it's very doubtful that the descriptions are complete. Look at transcript 3 again, and play the video if you can. Even at low resolution, you'll see many subtleties that are missing from the record. In fact, there is an infinity of things that could go into the description.

So where ought we stop?

The safest, though not always the most convenient, answer, is to work with the recording always available to hand.

But transcripts are essential tools. The written transcript, as I mentioned in the Introduction to this section, lets you work through things very painstakingly. So what do we put in?

The practice of most CA people is to stick with writing down what was said. They tend to keep descriptions to the minimum that the reader needs to make sense of who is addressing whom.

Most transcripts in CA are records of what was said. Of course, as we saw in transcript 2, 'what was said' doesn't just mean the words. We ought to include what people do with them in the way of pausing, drawing out, overlapping and so on.

We should then start doing whatever analysis we are up to, and see how far we get. If we have the video evidence to hand, well and good; we can supplement the text with the sound and / or vision. If not, then we bear in mind that certain things might come unstuck; but we have a go, and see how far we get.

A workable transcript

On the last page of this section (transcript 4), we'll return to a more 'wordy' transcript and explain the notation conventions.