**Warning - dull post follows that is purely for my own purposes**

I'm currently playing with the first transcript, trying to come up with a way to record information from it about places that are mentioned and talked about.  The categories are a combination of things that the env psychology literature flagged up as important in terms fo how poeple relate to space and place, stuff that seems obvious to us from a methodological point of view, and some more experimental ways to try and capture local history. At the moment I have the following categories:

Tag	Number of transcript, number of occurrence
Noise level	From transcript
Env descriptor	Type of area
Box number	10 second chunk
Speech object	Thing being spoken of
Type	F (factory), C (consumption building - e.g. shop / pub), P(public building - e.g. church, school), D (domestic), R (road), E (environmental feature), I (infrastructure), O (other)
Toponymy	G (general - no specific names), D (description using names), S (specific name)
Genesis	P (interviewer prompted), U (unprompted)
Presence	Y (still there with same use), D (still there with different use), N (not there)
Spatial descriptor	How its spatial location is described: R (road), B (building), S (sight), N (no spatial ref)
Location	Where the interviewee is in relation to the thing mentioned
Story	P (personal), I (impersonal) / b = built form, c = city, p = people
Topic
Preference	p (positive), n (negative)
Evaluation	p (positive), n (negative)
Place relation	d= elsewhere in digbeth, c= city centre, b= birmingham, o=other city)

Sorry about the formatting - that happens when you import excel stuff into web text editors, but you get the picture.  There's plenty of issues - for example, some places are mentioned more than once in the transcript. It's hard to get the balance between quantifiable typology and recording some qualitative aspects of the data.  I have gone for a half-way house, with between 3 and 6 tags for each variable.   I've also numbered the 10 second boxes in the transcripts so as to get a handle on the patterns in which places are mentioned.  It may also make it easier to categorise the overall transcripts in terms of discourse type.

Phil reckons these variables can be mapped quite easily by importing into a GIS, which should produce some interesting results.  It's literally spatialising discourse!  Cool.