First of all I tokenized transcript_resp tier via Tier > Tokenize Tier… and got transcript_resp_tokenized tier ([1], see the picture attached). Then it was exported as a tab-delimited text [2] and imported back [3], resulting in the new Imported tier. The problem is that when I either copy it or change parent of it so as it has transcript_resp as a parent and its stereotype is symbolic subdivision the order of annotations changes in comparison with transcript_resp_tokenized or imported tiers.
Hello,
Yes, this kind of copying/reparenting indeed leads to disappointing results. I think what happens, is that the annotations on the Imported tier are processed individually, one-by-one (because it is a top level tier). In this particular case it is very clear which annotations on Imported should be processed as a group and what the order should be, but that may not always be the case.
In the current implementation, for each annotation on Imported it is checked where it should be inserted, based on both its own start and end time and the “virtual” (interpolated) start and end times of the annotations that are already on the symbolic subdivision tier. Although I can imagine how this could be implemented differently (better?), this is how it currently is.
What you could do: first copy Imported as a time subdivision dependent tier of transcript_resp and then copy that new tier (Imported-cp) as a symbolic subdivision dependent tier of transcript_resp. The intermediate tier can then be deleted.
-Han
1 Like