How to unlinke multiple codes from multiple quotations at the same time

[SerraSerra]

Dear all,

I hope somebody can help with this issue.

So we started a team-project in Altas.ti , our PD's are transcripts of focus group discussions.

Each member of our team had been assigned the task to code each fragment where someone speaks with a speaker code and with some socio-demographics for every speaker we're interested in (gender, education, etc.). This was done by using the autocode function (look for 'speaker_name' , code with new code 'speaker_name') , then we duplicated/cloned the 'speaker_name' code for every socio-demographic we were interested in and renamed the clones/duplicates accordingly.

However, some in our team used the autocode function without checking the 'confirm matches' box. This is problematic since now every time a specific speaker is mentioned, (either because another speaker says his/her name, such as , either in descriptive statements based on our video-recordings such as ' Eva looks at Heinrich who is looking at his mobile phone and laughs while indicating him to John and Mary' ) it has been coded with both the speaker code and that speaker's socio-demographics. Obviously we don't want this and need to rework our coding so that only the fragments where Eva is speaking are coded with her speaker code (e.g. Speaker_Eva) and her socio-demographics (e.g. Gender_female , Education_Phd , Nationality_French, etc.). Luckily for us, the transcript is structured as follows:

Speakername utterance, blablablabla , (looks at Otherspeakername)

Otherspeakername response, blablabla

So I thought the easiest way would be to open the code manager and click a speaker whose codes I want to revise, eg Otherspeakername , by sorting the quotations I can then quite easily see all auto-coded fragments which have been wrongly attributed to Otherspeakername_code such as 'Speakername utterance, blablablabla , (looks at Otherspeakername) ' because they do not start with 'Otherspeakername'. However, this is where my difficulties start since: I can not choose to 'unlink selected fragments' from 'Otherspeakername_code' but can only select 'delete quotation' which would mean that in the case of the fragment : 'Speakername utterance, blablablabla , (looks at Otherspeakername) ' , it is also uncoded for Speakername_code ( + speakername's socio-demographics) instead of just for Otherspeakername_code (+ his sociodemographics). Moreover, even if I were able to unlink multiple fragments from one code in one instance, the problem would still be that fragments such as the above are still coded with both the socio-demographics of the actual speaker and those of the (uncoded , or to be uncoded) non-speaking but mentioned participant. Concerning how to batch uncode the wrongly coded socio-demographics , I haven't even considered how to resolve the issue yet.

So please if someone at Atlas.ti or on this forum has any suggestions in how to deal with the issue, it would be greatly appreciated. Many thanks in advance !

M. C.

[SFriese]

I think I already answered this in the helpdesk - but just in case someone else runs into this problem in the future - please do use unique speaker identifiers like speaker name: and then you search for speaker name: instead of just the speaker name.

Version 8.2 offers a new function to code speakers in focus groups - and it does require unique speaker names. So, I hope that something like the above issue will no longer happen.

Generally a good way to unlink lots of quotations from a code is to use the unlink option in the network. You can highlight all quotations that should be unlinked at it is one action to unlink them from a code.

I am not sure how easy this is in the above case - is it easy to identify all quotations easily? If so, you can highlight them all in the quotation manager, open them in a network, add the codes they should be unlinked from to the network as well and use the unlink option in the network.

Another option is to to delete all speaker codes and auto code them again. It does not take that long to redo it. This might be quicker than identifying all quotations that you do not want in the Quotation Manager.

If it only applies to one document or only a few - an option is to create a snapshot of the project. Remove all properly coded documents, delete all speaker codes from the snapshot and repeat the auto coding in the snapshot.

Remove all wrongly coded documents from the original project. Then merge the snapshot with the original project.

Susanne